Seichy\'s Missing Bit

Leica CL

Mon, 19 Jan 2026 00:00:00 +0000

Last year I got back into analog photography. As more and more of my devices seem to randomly turn on me, dumber devices have become surprisingly comforting. I've acquired a lot of analog cameras in the past year, as I try to find one that gets me. I wanna talk about these devices, and this Leica Cl is one of them.

{% image leica-cl.jpg "A picture of my Leica Cl" %}

The Leica CL is a 35mm compact rangefinder camera with interchangeable lenses in the Leica M-mount. It was developed in collaboration with Minolta who manufactured it. It first appeared in April 1973 and was released in the Japanese market in November 1973 as the Leitz Minolta CL. Both the Leica CL and Leitz Minolta CL were manufactured in a new Minolta factory in Osaka.

– Wikipedia

I am a sucker for old devices and ones with a good story behind them doubly so. The Leica CL is pretty close to being the perfect camera for me, it's a really compact and light camera, with terrific lens options and a great ecosystem around it. I particularly value its size, as it makes it a great everyday camera, I usually use a compact flash with it, it's a pretty great package and pretty comparable to my favorite digital camera, the Fuji X100v.

It can use most M-mount lenses and was originally sold with the Leitz Summicron-C 40mm f:2 and the Leitz Elmar-C 90mm f:4. Both are really compact lenses with great build quality. I haven't used the 90mm much, I really like the 40mm lens and well, I am a creature of habit.

I got quite lucky and got a full set, with both of it's original lenses in like-new condition for 900 euros. A fair price, I believe.

Here's a couple of my favorite pictures from the past couple of months, all taken with the Summicron-c 40mm F2:

{% image leica-cl_fuji-200_19.jpg "A picture of a water tower" %}

{% image leica-cl-portra-800_14.jpg "A picture of a snowy hill" %}

{% image leica-cl-portra-800_27.jpg "A picture of a carousel at night" %}

Additional Documents

Restoring my Adana H.S. 2 6 x 4 Letterpress

Fri, 11 Apr 2025 00:00:00 +0000

I've been meaning to get a small Adana Letterpress forever. I saw one a couple of years ago, and I could never get over it. Because of that, I've had an eBay search setup for years, waiting for the day the perfect specimen showed up. It finally did.

{% image 0cb403efbaf0a160f49093243152676e.jpg "My letterpress, in bright red with a little rust" %}

The Adana Agency was founded in 1922 by Donald Aspinwall in Twickenham. I find it extremely fascinating how it always seemed to be directed towards the hobby printer. The machines are reliable and easy to maintain. They have no bearings, most moving parts are metal on metal with only oil to ease their movement. They are still fairly popular and there's a decent amount of information online on how to take care of them. In 1987 the company was purchased by Caslon who still provides some parts for these machines today.

{% image fed822d2d79da6930b5832c60320a70b.jpg "An old advert for the Adana with the phrase 'Print it yourself' in big bold letters." %}

I got myself a small 6×4 model. The High Speed 2, as it seemed to be called, is a nicely sized machine. It is perfect for postcards and other small prints, it's also a tiny, lightweight machine which is perfect for my tiny, lightweight apartment.

It was in good shape, had some corrosion, but all pieces seemed to move and most importantly, be there. I wanted to get it done over a weekend, so I had to plan accordingly.

{% image 7db76ae2155d57f837cb7508007def58.jpg "Another angle for the letterpress, showcasing how rusty it is." %}

I started work early on Saturday, disassembling the entire machine and sorting the pieces that were rusty and the ones that were only dirty. This took quite long, a lot of nuts were rusty enough that removing them was challenging. I used a lot of bike chain oil (I like using it in lieu of penetrating oil) and waited, eventually most of them gave in and could be unscrewed. Luckily, the nuts that couldn't be removed didn't really prevent the rest of the machine from being disassembled. By the end of the day, I had all my pieces ready for cleaning.

{% image b7efc500d8f519d14c462bab5208d7d4.jpg "All the pieces laid out on the table before being cleaned" %}

On Sunday, everything that was rusty went into a bucket with Evaporust. I was doing most of this in my living room, so Evaporust was the only real rust-removing solution I'd be able to use safely. Everything was left in the solution and I checked it every 30 minutes, removing the pieces as they became rust-free. I didn't want to leave anything longer than what was absolutely needed because I worried about things like springs become brittle.

By the end of the day I had cleaned everything and I left all pieces out to air dry. I couldn't finish the project on time, so I had to do a couple of final things on Monday before final assembly. All moving parts that had rust removed received a spray of WD-40. This should prevent them from rusting right away. They would get some bike chain oil once the machine was assembled as well. I built a small plywood platform to mount the machine on, once it was fully assembled.

{% image 87e10e434908233f946a732e283bcfea.jpg "All the pieces laid out on the table after being cleaned" %}

Assembly was nice and quick, most pieces only fit in one way, and it's not hard to figure out¹. Once done I oiled all moving parts and filled in the small oil ports that cover the machine. I did this a couple of times until everything felt well oiled and no squeaks remained.

{% image 4a03e316d66cf9c67f9a1cf24781d120.jpg "The press put back together" %}

I proceeded to mount it on the board I had prepared, oil everything again and 3d printed a couple of holders for rollers and the ink disk. Finally, I was ready for a test run.

{% image 58e653a16f67719fb38ea9e29905ee27.jpg "The final press, mounted on a plywood board with a 3d printed block on the chase" %}

I don't have type. I also don't think I want to have type. I would love to have type, but I am generally speaking, scared of lead. If I managed to find non-lead alternative to type I'll use that. In the meantime, I am experimenting with 3D printing, my first attempt went well enough, and I think with a couple of small improvements it'll be good enough for the kind of things I want to print.

That's a lie. Oddly enough, I did put the roller arms backwards the first time, making it so everything looked like it worked but without a spinning ink disk. This took a bit to figure out, it wasn't until I checked everything again that I caught it. ↩︎

Bookbinding Adhesives

Tue, 14 Jan 2025 00:00:00 +0000

I've been doing a lot more bookbinding lately. I've done a lot of it throughout the years and recently decided to dive a bit deeper.

{% image notebook.jpeg "One of the notebooks I've made in the past month" %}

As part of this dive I've tried looking a bit deeper into the adhesives I use. I use Bookplast 22 most of the time. It's sold on my local bookbinding supply store, dries quickly and flexible. It is based of Methylisothiazolinone, which I've never heard of before. I think PVA is probably more commonly used though.

I've ruined a couple of notebooks because the glue I use dries too quickly. The common solution I've seen referenced is to use paste, which I am not a fan of. I've also started experimenting with methylcellulose, which seems like a common paste replacement. It's easy to make at home and unlike paste doesn't go bad (I haven't had any go bad yet anyways). It is extremely weak though, and sadly it only seems to work really well between papers.

Finally, the better alternative seems to be a mixture between the glue and methylcellulose. A 60 / 40 mixture is what's usually recommended, it gives you the strength of glue with the longer cure time.

I did a quick test where I pasted paper, bookbinding cloth and cardstock to greyboard, bookbinding cloth and cardstock to see how each behaved. I ripped it off after about 2 hours of cure time.

Here are the results:

{% video out-greyboard.MP4 %}

{% video out-cloth.MP4 %}

{% video out-cardstock.MP4 %}

Conclusion

I think mix is the way to go when I need to be precise, it dries basically just as strongly as glue while still giving me some freedom to move the pieces around if I don't get them perfectly straight. For everything else, there's glue. I want to try my hand at restoration and I think methylcellulose will have a place in my toolkit then.

We moved to Spain

Mon, 15 Jul 2024 00:00:00 +0000

2024 has been wild. Towards the end of 2023 we decided to move to Spain. I've always loved Spain and it's the country I've travelled the most to. We made the decision randomly one night and so the scramble started.

The first step was finding a lawyer and gathering documents. We knew we didn't need the added stress of DIYing it. We scheduled a couple of calls for the following week and were sold on the second one. They guided us through the process and we gathered all of the documents over the following month. The highlight was probably having to get married to get a marriage act. This was towards the end of October and we were only getting started.

We couldn't leave without our cats. Having four of them only made it more tedious. To import them into the EU we needed a clean rabies test from a certified lab. The closest lab was in the US and getting the results would take a month. The result isn't even valid until 3 months after the blood is extracted so that push our timeline to January at the earliest. We still needed to start the visa application process and pack up our apartment so we needed to get a lot going.

Angela left for Spain on the second week of November to get the application started and to get a new apartment. In the meantime I stayed back and started packing the apartment and getting the cats ready for the move.

Angela got a positive reply on the last week of December as I was almost done packing. With the papers now in order I had to figure out how to get the cats to Spain. Only one cat is allowed per passenger on the cabin so I had to get 3 friends to come with me to deliver the cats.

We left the third week of January, everyone got to Spain safely.

Project Log: MidiFaders - ESP-32 MIDI-BLE Faders

Sat, 04 Feb 2023 00:00:00 +0000

A while back I built a BLE MIDI controller. I was never too happy with it and had been meaning to take another stab at it.

My plan is pretty simple, I want a 4 or 5 faders in a small battery-powered box. I recently came across a TTGO display board. It brings most of what I need onboard, bluetooth, a screen, a couple of buttons, battery management and, best of all, a usb-c port.

I knew exactly what I wanted, so I went to my parts cabinet and grabbed some faders, and measured how many I could fit easily on that board.

Electrical Design

The electronics were pretty straightforward. I just needed all of my faders to be connected:

{% image schematic.svg "A schematic for the midi faders" invert %}

I rapidly routed a board, and proceeded to mill it.

Milling a PCB

I have a small desktop CNC I use for milling PCBs at home. I don't use it nearly as often as I'd like but I love having it available. It's much better than protoboard and way more fun.

I had a lot of issues milling this PCB though. It turns out most of the software I am used to using in Ubuntu flat out doesn't work on a Mac M1. So I had to find solutions to some of those issues.

The biggest problem was FlatCam. FlatCam is awesome and I love it but it seems barely maintained and running it is getting harder every day. I decided to try out alternatives. It turns out there really aren't any.

I ended up settling with Copper. This is a website provided by carbide3d that allows you to upload your layers and get a set of gcode files. You only get minimal options so I don't think I'd use it for anything but a one-sided PCB but it seemed acceptable.

{% image 2.jpg "A picture of Carbide Copper" %}

With my set of gcode files I felt ready to mill my PCB. Thankfully bCNC still runs flawlessly on a Pi so that part of my workflow remained intact. Here's a quick rundown on how I usually do it (I'll probably write a separate post to go into it in more detail).

I start by prepping a PCB blank. I clean it will alcohol and a scotch pad and set it on the bed using double-sided tape.

Next, I probe the PCB and set my working space.

{% image 3.jpg "The spindle probing the pcb blank" %}

I the autolevel but don't click autolevel. If you've used bCNC before you know this breaks stuff.

{% image 4.jpg "The resulting PCB." %}

After cleaning it up, we can do soldering. This is generally the easy part. I leave as much space as possible around my traces, as much as 1mm if there's space. Soldering comes down to a lot of flux and as much solder as it takes. I'll usually coat parts of the PCB with the excess solder, this is probably useless but I feel like it makes for a better connection on these homemade PCBs.

{% image 5.jpg "The parts inserted into the PCB before soldering." %}

I'll do a quick fit check with all parts and then start soldering.

{% image 6.jpg "The PCB now soldered." %}

It ended up looking pretty good.

{% image 7.jpg "The PCB from the frontside." %}

Firmware~

So the firmware for these things is usually the hard part. My idea was extremely simple. Each fader should send, via bluetooth midi, a cc on a channel. These should be modifiable from within the device. The UI would have to be something I came up with as I went. I used the excellent arduino_midi_library with a bluetooth transport and everything went surprisingly smoothly.

In a couple of days I had implemented all menus, battery management, bluetooth and settings. Of the features I wanted to add, I am only missing presets and a web app to load these in. Overall I am extremely happy with how performant and easy to use it is.

Case

The case took a couple of iterations to get right. I hadn't done any 3d printing in a while so I was a bit rusty.

{% image 8.jpg "A 3d printer, printing the case" %}

After shifting around the design a bit I came up with something that was both easy to print and very functional.

{% image 9.jpg "The different versions of the case" %}

The End ~

I am very happy with the final result. It came out really clean, easy to use and highly functional. Setting up new devices is fast and easy and since it's compliant it connects seamlessly with other MIDI-BLE devices, like the widi masters.

{% image 10.jpg "The final device" %}

{% image 11.jpg "The UI showing the midi channel selector" %}

{% image 12.jpg "Another closeup of the UI" %}

Making Softcases for Small Electronics and Cables

Sun, 29 Jan 2023 00:00:00 +0000

Wraps are my favorite way of storing small electronics. Whenever I get a new one I'll probably make one really quickly to store it and it's accessories.

It's a fairly straightforward process. Start with a piece of cloth. I really like this one, where it's like felt on one side and like linen on the other. It makes it so the interior is really nice and soft.

{% image 1.jpg "A large piece of fabric laying on the floor" %}

I'll start by measuring the device I'll be wrapping. Once I've got the measurements I'll lay my cloth on the floor and draw a rough rectangle. It'll be a bit wider than needed and about four times the length of the device. This will let me wrap it comfortably a couple of times.

{% image 2.jpg "A couple of lines chalked up on black felt" %}

I'll then cut this rectangle off.

{% image 3.jpg "A small piece of fabric laying on the floor now" %}

I'll then fold a third of this cloth on top of itself and sew it. I got a free sewing machine a while back (all it took was a couple of 3d printed replacement parts to get it working but that's a story for another day) which I use for this.

{% image 4.jpg "A sewing machine... sewing" %}

{% image 5.jpg "A close up of the sewn case" %}

Once that's done I'll flip it inside-out and sew any open borders to give it a cleaner look.

I'll also sew a elastic on the open side so I can close it up.

{% image 6.jpg "The Elastic sewn to the flap" %}

I am not great a sewing.

I'll then place the device inside, wrap it up and figure out where to put a button so that it closes up cleanly.

{% image 7.jpg "A button being placed." %}

Buttons are hard. I'll do as many passes as possible because I really suck at place them. Just doing a lot of passes will usually yield a nice looking result.

{% image 8.jpg "A closeup of the button sewn into place." %}

{% image 9.jpg "How it looks on the insider" %}

{% image 10.jpg "How it looks on the outside" %}

It's all about confidence. Once it's all done, you can insert the device in and wrap it, using the elastic as a latch.

{% image 11.jpg "A closeup of the button sewn into place." %}

DIY Midi Cable for the OP-Z and ZM-1

Sun, 15 Jan 2023 00:00:00 +0000

I've been using an old Rockband Keytar as a midi controller for the longest time. I usually take advantage of the OP-Z's ability to be a MIDI host to use the keytar via a midi to usb adapter. This is fairly cumbersome as I have to use big cables and a USB-A to USB-C adapter.

Last year I got myself a ZM-1 module for the OP-Z which allows me to connect MIDI devices via a 3.5mm jack. These are fairly easy to put together and I had all the spare parts I needed hanging around already so I decided to make a couple.

All you need is a MIDI jack, and 3.5mm jack and some cable. I used a midi jack that already had it's cable attached. I took this from a previous midi project so that cut the soldering in half.

{% image 1.jpg "The parts we'll need. A midi jack and a 3.5mm jack"%}

I couldn't find documentation stating weather this cable is meant to be type a or b or something else. So I tried wiring it as a type-A (a being the "standard" from what I know). This didn't work, it turns out is actually type-B. Here's the diagram:

{% image diagram.svg "The diagram we'll follow, current source to tip, current sink to the ring and shield to the sleeve "%}

Some soldering later and the cable is mostly ready:

{% image 2.jpg "The cables soldered onto the 3.5mm jack"%}

I used some heatshrink as strain-relief. I was also lacking the piece of plastic that screws to the 3.5mm jack so I wrapped it in some extra heatshrink to give it some strength. Here's the final cable:

{% image 4.jpg "The finished cable, now with some heatshrink" %}

Here's what my keytar setup looks like now:

{% image 5.jpg "The op-z on top of my rockband keytar" %}

It works flawlessly and makes for a fun little weekend project.

Project Log: Some Sort of Guitar

Mon, 02 Jan 2023 00:00:00 +0000

I was lucky enough to be gifted an old, broken guitar.

{% image 1.jpg "An image of the guitar's body" %}

It was in pretty rough shape. The bridge had been bolted to the body so when it started lifting it broke through the top.

{% image 2.jpg "A closeup of the guitar's lifted bridge" %}

I felt that it wasn't really worth restoring it and that I'd rather try building something new on top of it. I was thinking that just adding a pickup would make it a bit more interesting.

I started by removing half of the top. Since I don't care too much about the quality, I was able to remove the lower half quite easily with a utility knife.

{% image 3.jpg "A ruler on the guitar showing where I cut it through" %}

{% image 4.jpg "The lower half of the guitar's top removed from the guitar" %}

I used this lower half to create a template to create a new lower half.

{% image 5.jpg "The paper template I created"%}

I used this template to cut a thin sheet of plywood to replace that piece. It's a bit oversized but I will be sanding it down to size.

{% image 6.jpg "The guitar with its new plywood top"%}

The new plywood was a bit thicker than the rest of the top so I had to plane the outer wall down so that both parts would be somewhat flush together.

January 22nd

Here's the not-so-fun part. I started prepping the guitar for painting. I am thinking I want to paint the entire thing in a solid color, a dark one should make the entire thing easier to match.

I started with the body. I needed to make sure all edges were as neat as possible. There's an upper bound to how good I think I can get it to look so I am trying to cover the basics at least.

I didn't have a chisel on hand so I had to use a utility knife. It made the process rather slow but I got it done.

{% video video-1.mp4 %}

After cleaning the edges, I sanded the body.

{% image 7.jpg "The guitar with is freshly sanded body"%}

Next I had to do the neck and fittings. I started by removing the old fittings, I don't intend to buy new ones so I'll have to clean these.

{% image 8.jpg "The guitar's tuners"%}

Luckily disassembly is pretty straightforward.

{% image 9.jpg "The guitar's tuners dissambled"%}

Next I prepped the neck. Some light sanding...

{% image 10.jpg "The guitar's neck after a light sanding"%}

and some light scrapping...

{% image 11.jpg "The guitar's neck after a light scrapping"%}

and it was ready to be painted.

{% image 12.jpg "The guitar hanged from a wire to be painted"%}

We applied a primer, to hopefully even everything out a bit.

{% image 13.jpg "The guitar with it"s white primer'%}

Now it only needs a second coat of primer and we'll apply it's final color.

January 24

In the meantime, I've gone ahead and cleaned up it's original hardware. I left it all submerged in a cup with 1 part of distilled white vinegar and 1 part water overnight.

The next day I had a go at it with a scotch pad and some steel wool under running water.

{% image 14.jpg "All the parts cleaned up"%}

After letting it dry and having it reassembled, it all looked great.

{% image 15.jpg "The tuners reassembled"%}

Installing Gitea on Dokku

Sun, 07 Aug 2022 00:00:00 +0000

I've been meaning to backup my GitHub repos somewhere for a while. I recently discovered that Gitea can mirror them and will keep them updated. I am running Dokku on my dedicated server and pretty much run everything I can think of there. Gitea was a bit tricky to install so here are some notes.

You start by creating the app:

; dokku app:create gitea

Once that's done, we can initialize it from the docker image:

; dokku git:from-image gitea gitea/gitea:latest

This is a good time to set the correct domain name and mount the Gitea directories.

; dokku domains:add gitea git.example.org
; dokku storage:ensure-directory gitea
; dokku storage:mount gitea /var/lib/dokku/data/storage/gitea:/data

We can rebuild the app now:

; dokku ps:rebuild gitea

Dokku will get some ports wrong sometimes, we should check those:

; dokku proxy:report gitea
=====> gitea proxy information
       Proxy enabled:                 true
       Proxy port map:                http:22:22 http:3000:3000
       Proxy type:                    nginx

Those look somewhat wrong, lets fix them:

; dokku proxy:ports-remove gitea http:3000:3000 http:22:22
; dokku proxy:ports-add gitea http:80:3000

We can enable LetsEncrypt now:

; dokku letsencrypt:enable gitea

That's pretty much it! Seems a lot simpler once I've written it down. If you want to change the config file you can go ahead and edit it + restart the app:

; sudo vim /var/lib/dokku/data/storage/gitea/gitea/conf/app.ini
; dokku ps:restart gitea

Cheers!

The Pandemic

Fri, 05 Aug 2022 00:00:00 +0000

As the end¹ of the Pandemic approaches, I wanted to write a blog post to remind myself of these past 18 months and how we spent our time in exile. There isn't really a point to this blog post in general it's mostly a slice of my life and a lot of disconnected thoughts.

I originally started writing this post in August 8th, 2021. It took me a year to get around to finishing it but here it is. Better late than never I guess!

Before The Pandemic

My exile started a bit before the pandemic got to Honduras. Towards the end of the previous year (2019) I had a cancer scare. It turned out to be nothing but I developed a lot of anxiety around my health because of it. I had become overly weary of anything health related and hearing of this SARS-like disease that was showing up everywhere didn't help. By that time everyone was a bit tired of my shenanigans and when I asked everyone around me to prepare for it I (understandably) was not taken very seriously.

Honduras is poor and I knew it wouldn't take much for it's system to collapse. If even rich countries seem to be struggling it wouldn't really take much for the same to happen here. I told my mother that if she wanted to visit she'd have to do so before the first case was reported in Honduras. Once that happened I'd pretty much lock myself up and isolate for however long it took. She came to visit a couple of weeks before the first case, just in-case. It sounded like masks and hand sanitizer would be the first thing to fly off the shelves so I purchased some preemptively. I figured I was being overly dramatic but better safe than sorry. Having basic things I could hold to would make my anxiety easier to control I figured.

As the pandemic seemed to get closer I told everyone my plans of staying indoors until it was under control, that it might be a bit long but hopefully not more than a couple of months. The 8th of March, we had Arleth over for brunch, we talked about the upcoming pandemic over coffee.

On 10th of March, 2020 the first two cases of covid were reported in Honduras my lock down would begin two days later. The following day I dismissed the house staff, had them take the day off and go purchase groceries and supplies. I assured them that they would still receive their salaries and that they would have job once this was over.

The first month

Once the first cases were reported everything started moving too fast. By Sunday every supermarket was swamped with people and supplies started dwindling. I felt a bit validated knowing I had prepared myself with a couple months worth of food. That sense of validation didn't really last long.

For the first couple of weeks everything was under strict lockdown and looked like Honduras wouldn't spiral too bad. Wishful thinking.

The 20th we had a nice break with the release of Animal Crossing: New Horizons. It came in at the best time, the boredom was settling in and everyone with a switch was playing it. It felt like a community event, everyone was alone, playing together. It carried us over a month or two.

By the end of march the food was a little less varied so it seemed like the right time to resupply. It looked like all supermarkets were booked for deliveries for a month or so in advanced so we had to get creative. For non-perishable we used a small supermarket that specialized in organic goods. It's variety wasn't great and it was really expensive but it allowed us to get some basics and some not-so-basics while the regular supermarkets got their act together.

Mid April everything got a lot better. We found a local catering company that being unable to cater, had started selling vegetables, dairy and eggs. Getting fresh vegetables for the first time after almost two months gave me back my will to live.

The following 4 months

As the pandemic raged on I realized I was not doing too great. I had become a bit of a hypochondriac and had developed some anxiety. I had been stuck in my apartment for a while now so it seemed justified. One day I decided I wanted to get into therapy. I found a nice Colombian psychologist that would treat me online. It worked out great and after a month or two I felt better and was told I could stop going.

Whenever I am bored I get started on new hobbies. I had a ton of paint around the apartment so I started painting. I would experiment quite a bit with acrylics, I didn't like how glossy they were but it grew on me. I eventually started drawing over them with markers and developed a bit of a "style". I would paint quite a bit over the next couple of months.

{% image painting.jpg "One of the many paintings I did over the pandemic. An infinity net over a red background."%}

As the sense of normalcy grew and I had more time off I started playing Call of Duty with Carlos and Carla. It happened almost by accident but we started playing a couple of hours every day. It was great fun, they've been my best friends for the longest time and being able to spend time with them, even if only online, felt awesome.

Eventually I fell back to my oldest habit, overworking. I decided I wanted to be more productive and get more done. I had been toying with an idea to help local companies get their products online. Honduras is really far behind on this sort of thing and during the pandemic it really showed. I've had an Estonian e-residency for a while so I used that to incorporate a business and started working on the app. I would let it sit to focus on my day job soon after.

Up to this point we had been using a small refrigerator I bought back when I first moved in. It was fine at the beginning but since getting delivery was hard it made more sense to order more food and store it than try to order when needed. We got a new fridge which was a pretty large QOL upgrade at the time. Sounds weird but it really cheered us up.

At this point my anxiety had improved so I decided to let myself drink coffee again. It's always been a large part of my life so it felt like that'd help me go back "normal". I decided to make a hobby off it (it technically already was) and got myself a espresso machine. I named it Pino.

{% image pino.jpg "Pino the espresso machine"%}

I eventually start working on my small side project, a site to simplify selling online, elmarket.

The rest of 2020

Very little happened the rest of the year:

We raised moths.

{% image marimoth.jpg "One of the moths we raised, marimoth"%}

We would raise more moths the following year at around the same time of the year.

Customs eventually reopened so we bought some extra art supplies. I got a new camera (Which I called Prismo). I painted some more. My sister suggested we run a book club, we read one book, tiny fire everywhere, I hated it and decided to stop book clubbing.

{% image painting2.jpg "Another painting of mine from that time"%}

I took a couple of weeks off in December to try to chill for a bit.

{% image sky.jpg "A picture of the sky"%}

A New Year

On my first day back after the Christmas break we were told at Azlo that our CEO was leaving. The following day they announced we would be shutting down. Our parent company had been sold off and no one really wanted to keep operating Azlo.

Since I was still stuck home I immediately started interviewing and looking for a new gig. I found a really nice job at Blueonion Labs. They were doing really cool things that closely matched the reconciliation stuff I was doing at Azlo. I would start the week after my birthday. I decided to ignore it that year, it didn't seem fair to count it.

We got a roomba shortly after, it was surprisingly helpful, we were spending quite a bit of time cleaning and just having it run everyday made it way easier.

I started working at Blueonion Labs.

We found Pitt, the chunkiest boy. We originally offered him to Edward for adoption but he didn't want to adopt him. After a couple of days, even though we thought we had enough cats, we ended up falling in love with him.

{% image pitt.jpg "Pitt, the cat"%}

We finally got our first vaccine shot in August. It was nice, it felt like maybe it was coming to an end.

A bit later my grandfather died from Covid. My uncle followed a couple of weeks later.

I left my job. I was feeling a bit burnt out and depressed.

Towards the end of September we were fully immunized and decided it was time to start moving on. We started planning a trip, it'd be the third time leaving the house in 18 months. The other two were for vaccinations.

Shortly before leaving we found Gomita, the newest member of our family. We had no intentions of adopting another cat at the point but she needed a ton of help and it didn't seem fair to pin her off on someone else. So we kept her. No regrets.

{% image gomita.jpg "Gomita, the cat"%}

We left for Europe in October and spent a couple of weeks over there. I had to go to Estonia to open a bank account for the company I had founded the year before but we spent the rest of time relaxing and doing normally touristy stuff (mostly eating).

{% image me.jpg "A picture of the author"%}

We are about 4 weeks away from being fully immunized which is as far as I know, the closest thing to an end there can be for the time being. ↩︎

Installing FlatCAM on Ubuntu

Fri, 22 Jan 2021 00:00:00 +0000

So FlatCAM is an essential part of my PCB milling workflow and has become increasingly difficult to install on the latest version of Ubuntu. Here's a couple of notes on how I managed to get it running.

The first thing is to clone the repo:

git clone https://bitbucket.org/jpcgt/flatcam

Something I didn't notice at first is that there's a beta branch available, it uses newer versions of pyqt which makes everything way easier.

git checkout origin/Beta

If you are on Ubuntu you can run ubuntu_setup.sh and should install pretty much everything you need. I did hit a snag though, for whatever reason if I tried installing the requirements, gdal would not compile. Turns out it comes down to it not finding the correct headers. You can get around it by installing it manually like this:

sudo apt install libgdal-dev gdal-bin
pip install gdal==3.0.4 --global-option=build_ext --global-option="-I/usr/include/gdal/"

You should use whatever version gdal-config --version prints out. You should also update the requirements.txt so that the version of gdal is also the same so that the rest of the dependencies install.

Once that's up you can run it with python FlatCAM.py.

Haproxy, Nginx and Dokku

Mon, 04 Jan 2021 00:00:00 +0000

I just finished setting up a wordpress site for my sister over the past weekend. I usually setup some plugins to automatically block bruteforce attempts on these since they are so common. Everything went smooth as usual until I checked in on it the next morning and found myself blocked. I checked the logs and it turns out it banned the reverse proxy's ip.

I am using Haproxy as a reverse proxy / load balancer for the entire server but the ssl is handled by dokku which means Haproxy runs in tcp mode and doesn't do much other than route the traffic. Dokku also has a reverse proxy which it uses to handle domains, it uses Nginx for this.

It turns out Haproxy can't add requests to an https request that terminates on a different server. This makes sense but wasn't obvious from the get go, the easy solution was to use the proxy protocol so that Haproxy and Nginx can agree on which ip is which.

Here's how to do it:

First, we need to tell Haproxy to use the proxy protocol. All we have to do is add send-proxy to the server option. This will send over all of the info Nginx will need to find the correct ip.

backend web
        mode tcp
        server web-http dokku send-proxy

Next we need to tell Nginx to actually use that info. This is done by adding the proxy_protocol parameter to the listen option.

server {
  listen      443 ssl proxy_protocol http2;
  ...
}

Finally, we need to set the request's X-Forwarded-For header to the real ip. We can achieve this by adding a new conf file in /etc/nginx/conf.d/set_real_ip.conf. You can name it whatever you want, it should look something like this:

set_real_ip_from  10.0.0.0/8;
set_real_ip_from  172.0.0.0/8;
real_ip_header proxy_protocol;

You should list off all of the proxies that are part of the chain here. proxy_protocol will take care to set the X-Forwarded-For ip to the correct value automatically.

KVM Backups, the easy way

Fri, 30 Oct 2020 00:00:00 +0000

I am now using a dedicated server to host all of my various projects. It turns out if you are willing to do a lot of the work yourself you can get quite a bit for you buck. Since I have to take care of backups myself now I had to rethink my general backup strategy.

In the name of being as lazy as possible what I've been doing now is using libvirt to create individual VMs for every project. To back them up I can just copy off the entire disk somewhere else. Once compressed these are relatively small for what they are and are easy to move around into storage.

I am also using git underneath almost everything so that actual code and assets are also backed up locally and on github/gitea etc.

Anyways, here's what I am using to back machines up:

DOMAIN="$1"

DISK="$(virsh domblklist $DOMAIN | tail -n +3 | awk '{print $2}' | tr -d '\n')";
TIME=$(date "+%Y-%m-%d--%H.%M.%S");

virsh snapshot-create-as --domain $DOMAIN "$DOMAIN-backup" \
        --diskspec vda,file=/var/lib/libvirt/images/$DOMAIN-backup.qcow2 \
        --disk-only --atomic --quiesce --no-metadata;

xz -k --verbose /var/lib/libvirt/images/$DOMAIN.qcow2;

curl --netrc -T /var/lib/libvirt/images/$DOMAIN.qcow2.xz ftp://ftp.example.org/$DOMAIN.$TIME.qcow2.xz;

virsh blockcommit $DOMAIN vda --active --verbose --pivot;
rm /var/lib/libvirt/images/$DOMAIN.qcow2.xz;
rm /var/lib/libvirt/images/$DOMAIN-backup.qcow2;

First I create a snapshot for the current machine I'll be backing up, these machines are running the qemu agent so I get to use --quiesce to freeze the file system. If you don't have the agent installed you'd probably have to make some small changes to the command.

virsh snapshot-create-as --domain $DOMAIN "$DOMAIN-backup" \
        --diskspec vda,file=/var/lib/libvirt/images/$DOMAIN-backup.qcow2 \
        --disk-only --atomic --quiesce --no-metadata;

Once that's done I'll compress it using xz. This will make the backup tiny but it takes a really long time, you could use zip, tar or nothing at all if you want the whole process to be faster.

xz -k --verbose /var/lib/libvirt/images/$DOMAIN.qcow2;

Now we can upload the compressed file to our backup location. I use curl to upload to a local network drive I have available. I also attach a timestamp to the file name as I do this.

curl --netrc -T /var/lib/libvirt/images/$DOMAIN.qcow2.xz ftp://ftp.example.org/$DOMAIN.$TIME.qcow2.xz;

Finally I will use blockcommit to pivot back to the original disk image, merging the contents of the backup and anything that's happened since the snapshot was taken, and remove the unnecessary snapshot and compressed files.

virsh blockcommit $DOMAIN vda --active --verbose --pivot;
rm /var/lib/libvirt/images/$DOMAIN.qcow2.xz;
rm /var/lib/libvirt/images/$DOMAIN-backup.qcow2;

It took me a bit to figure out how to do this which is why I am documenting it here for myself. Most of this is based of off this wiki entry.

PocketChip in 2020

Sat, 24 Oct 2020 00:00:00 +0000

After many years of wanting one and then giving up on them when the company went under, I finally acquired a PocketChip. This thing is great. Not all around great but great in principle. It has a great form factor with a nice screen and great battery life. I quite like it. I've been using it to control my plotters wirelessly, for counting words on my typewriter and to backup my op-z. I've ran into a bunch of issues along the way so here are my notes on it, what's worked and what hasn't in case someone else wants to give it a go.

Apt Sources

The first thing you should do is change you apt sources. It turns out when the NextThingCo went under they took their mirrors with them (naturally). Luckily a valuable member of the community took the time to create a mirror for it we can all use here. That same member of the community also took the time to mirror all of NextThingCo's repositories and other goodies: http://chip.jfpossibilities.com/ this is in general the single best resource of data and code available for the chip in 2020.

I ran into an issue with Jessie backports, as the distribution is pretty old it'll be necessary to update your apt sources to use the debian archive mirror:

deb http://security.debian.org/ jessie/updates main
deb-src http://security.debian.org/ jessie/updates main

deb http://archive.debian.org/debian/ jessie-backports main
deb-src http://archive.debian.org/debian/ jessie-backports main

deb http://archive.debian.org/debian/ jessie main contrib non-free
deb-src http://archive.debian.org/debian/ jessie main contrib non-free

Don't forget to disable validity checks:

echo "Acquire::Check-Valid-Until false;" | sudo tee -a /etc/apt/apt.conf.d/10-nocheckvalid

SSH Access

The PocketChip doesn't include an ssh server by default so I'd recommend installing one once you've gotten your apt updating and running:

sudo apt-get install openssh-server

Checking the Battery Level

There's two magic files in the os that reflect your current battery state, you can use /usr/lib/pocketchip-batt/voltage to check how charged your PocketChip is and you can use /usr/lib/pocketchip-batt/charging to check if it's charging or not. These are mostly useful for scripts and such.

The voltage range is from 3600 to 4200 from what I've seen.

Using UART

If you want to use the built-in UART you need to shutdown getty:

sudo systemctl stop serial-getty@ttyS0.service

You can then use /dev/ttyS0 as you normally would as a terminal or programmer and stty to check the config stty -F /dev/ttyS0 or change it stty -F /dev/ttyUSB0 9600.

A full desktop environment

If you want to have a full desktop environment running on your chip, you can use PocketDesk, which sets it up for you.

Using your GPIO Pins

This actually took a lot of tinkering and was going to be the original focus of this blog post.

PocketOS provides an easy way (once you figure it out) to enable or disable gpio pins. Under /sys/class/gpio there's two sinks export and unexport. As root you can use export to enable a gpio pin by piping it the pin's identifier:

echo 1015 > /sys/class/gpio/export

Apparently these identifiers changed depending on the OS version, on mine it starts with GPIO 1 as 1015 and follows with GPIO 2 on 1016 and so on. When you are done using a pin you should unexport it:

echo 1015 > /sys/class/gpio/unexport

While the pin is exported you can set three settings according to what you are trying to achieve:

Pin direction:

echo "in" > /sys/class/gpio/gpio1015/direction
echo "out" > /sys/class/gpio/gpio1015/direction

Active Low:

echo 1 > /sys/class/gpio/gpio1015/active_low
echo 0 > /sys/class/gpio/gpio1015/active_low

Edge detection:

echo "rising" > /sys/class/gpio/gpio1015/edge
echo "falling" > /sys/class/gpio/gpio1015/edge
echo "none" > /sys/class/gpio/gpio1015/edge

You can check the GPIO's state by reading the value:

cat /sys/class/gpio/gpio1015/value
# Either 0 or 1 depending on the current state of the pin.

Since this behaves like a normal file and emits events you can use anything that uses inotify to watch it. For example, here's how you'd monitor a button press using inotify-hookable:

cd /sys/class/gpio
echo 1015 > export
cd gpio1015

echo 0 > active_low
echo "in" > direction
echo "rising" > edge

inotify-hookable -f ./value -c "cat ./value"

There's some extra information about the pins in the community wiki. The pin numbers seem wrong for the pocketchip but the rest of the information available seems solid.

Configure the app launcher

The default built-in launcher for the PocketChip, PocketHome, can be minimally configured by editing a json file /usr/share/pocket-home/config.json. You can change the apps and the icons from there among other small niceties.

I'd recommend changing the terminal to xfce-terminal, it comes pre-installed and seems to be better in general, I changed the terminal button to this:

{
  "name": "Terminal",
  "shell": "xfce-terminal --hide-borders --fullscreen --hide-menubar",
  "icon": "appIcons/terminal.png"
}

You need to restart the entire PocketChip for the changes to kick in.

Also, if you are ever using a keyboard, dmenu is installed and mapped to alt + space which is super nice for quickly switching apps.

And that's it, I think, everything I've had to figure out about the PocketChip, distilled into one blog post. Let me know if there's anything important you think I should know.

Setting up Ruby on Rails System Tests on WSL with Docker

Tue, 26 May 2020 00:00:00 +0000

For some context, I recently started a new RoR project on Windows. The process went remarkably smooth. Everything worked out of the box. Well, everything except for system tests.

It turns out Capybara will, by default, any chrome instance installed on your local machine. That's normally great except for when there isn't one available. Luckily there is a surprisingly easy way to achieve this in a cross-platform way. Docker. There are great docker images that can run a headless Chrome or Firefox instance for you, selenium even has a hub to administrate all of these instances. The configuration for this within rails is easy but poorly documented which brings us to this post.

To get started you should setup Selenium hub and a browser or two, if you are using docker already it's as simple as creating this docker-compose.yml file:

version: '3.1'
services:
  hub:
    image: selenium/hub:3.141.59-20200515
    ports: 
      - '4444:4444'

  chrome:
    image: selenium/node-chrome:3.141.59-20200515
    volumes:
      - /dev/shm:/dev/shm
    depends_on:
      - hub
    environment:
      HUB_HOST: hub
      HUB_PORT: 4444

Once that's setup and running we can now go ahead and tell rails to use this hub for all of our testing needs. The first thing we have to do is remove webdrivers from our gemfile. As long as that's installed capybara will keep trying to use our local chrome instead.

Now we have to change the default chrome driver to use the selenium hub instead:

# application_system_test_case.rb 
require "test_helper"

class ApplicationSystemTestCase < ActionDispatch::SystemTestCase
  driven_by :selenium, using: :chrome, screen_size: [1600, 1080], options: {
    url: "http://localhost:4444/wd/hub",
    desired_capabilities: Selenium::WebDriver::Remote::Capabilities.chrome(
      chromeOptions: { args: %w[headless window-size=1600x1080] },
    )
  }
end

This would be mostly it if we are running rails containerized as well but since my rails runs on the host machine we need to tell headless chrome to visit rails on the host machine so you also have to modify the setup code:

# application_system_test_case.rb 
require "test_helper"

class ApplicationSystemTestCase < ActionDispatch::SystemTestCase
  ...

 def setup
   Capybara.server_host = "0.0.0.0"
   Capybara.server = :puma, { Threads: "1:1" }
   Capybara.app_host = "http://host.docker.internal:#{Capybara.current_session.server.port}"
   host! "http://host.docker.internal:#{Capybara.current_session.server.port}"
   super
 end
end

The key idea here is that rails instances created by capybara when testing should be bound to 0.0.0.0 so they are available to the chrome container, these will get a random port assigned and we can use it along with host.docker.internal to allow chrome to connect to it.

It turns out it's pretty easy to setup once you figure it out!

Notes on a Roland DXY-1150

Fri, 25 Oct 2019 00:00:00 +0000

I recently acquired a second plotter to complement my HP-7475A. I've been meaning to get an XY plotter so that I can use taller pens without modifying the case of my hp plotter. These Roland plotters are both beautiful and very functional so I've been waiting for a nice looking one to show up on my feed. It finally did.

The Power Supply

I overpaid quite a bit for it but it was what I wanted so I figured it was worth it. It was about $150.00 without a power supply. A word of warning, DO NOT BUY ONE OF THESE WITHOUT A POWERSUPPLY. I am not the first person to say this but I decided to ignore that advice. I was lucky to find a power supply for it for an additional $30. All in all I ended up paying a bit over $200 for it (don't get me started on how much it was to ship it to Honduras). If you are looking for a power supply the model number is DCP-302A. It is a dual voltage power supply with one rail providing 31V 0.7A DC and another providing 9.7V 0.7A DC.

Here's the pinout just in case:

{% image roland-DCP-302A-pinout.svg "A diagram of the power supplys pinout" diagram %}

Communicating With The Plotter

Luckily this plotter has two different ports, which makes communicating with it very convenient. The serial port is fairly straightforward to use so I went with it. Turns out getting a USB to Serial adapter in Honduras was a bit harder than I expected. I did end up finding a DB9 USB to Serial adapter. The plotter has a DB25 serial port so that was only half of it. Luckily my plotter came with a PS/2 to DB25 cable I could sacrifice to make an adapter.

My DB9 USB to Serial adapter was male so I bought one of these DB-9 female adapters to replace the PS/2 end of my cable with.

Luckily Roland tells us what pin goes where in their user manual so setting it up was fairly easy.

{% image serial-interface-diagram.svg "Rolands pinout chart as shown in the user manual" diagram %}

Here's how I wired it:

DB25 Pin	DB9 Pin
FG (1)	Sleeve (GND)
TXD (2)	RXD (2)
RXD (3)	TXD (3)
DTR (20)	DTR (4)
SG (7)	GND (5)

Make sure that the TXD goes into the RXD and the RXD into the TXD. It might seem obvious but I got it wrong the first time. Luckily these solder-less DB9 plugs are easy to reconfigure.

That was pretty much it, I plugged everything in and it all came together.

DIP Switches

The dip switches are fairly straight forward, I just kept everything off. So just change everything as you need it. I'd recommend changing one at a time. Coolterm managed to connect to it without any special settings. Baudrate 9600, Databits 8, Parity None, Stop Bits 1, DTR and RTS ON.

Conclusions

This plotter was surprisingly easy to setup. Its worked almost entirely out of the box. Had I bought the correct adapter from the start it would have been a 5 minute affair rather than an hour long one.

Glitch, The Cat

Fri, 06 Sep 2019 00:00:00 +0000

Glitch the cat was my best friend for quite a while. This might seem as hyperbole, but that's exactly who he was. I don't think I've ever gone through anything as hard as losing him. It gets to the point that sometimes I wake up and think about how dull a world without Glitch is and how I might not want to live in that world. I carry on though. He was a beautiful and strong cat who had his own way of showing how much he cared for us. That's how I'll always remember him.

Goodbye Glitch, The Cat. I love you and you will be missed.

Restoring my HP7475A to it’s former glory

Fri, 26 Oct 2018 00:00:00 +0000

Recently I’ve become obsessed with generative art. As usual I dove in head first. I love old HP gear so seeing the HP 7475a as one of the most recommended plotters around made it a no brainer. Everyone seems to recommend avoiding getting one with a HPIB port; This makes them a lot cheaper than their serial counterparts. I managed to get one for 10 bucks!

A couple of weeks later it came in through the door. It’s a beautiful machine. It’s big and built like a tank, it feels ridiculous. I love this machine. It came with a couple of missing bits, the pen carousel was broken, it runs on top of a Geneva drive and the small gear was broken (it actually pulverized when I first tried moving it around).

Luckily Geneva drives are well documented and this one was a textbook example so designing and 3D printing a replacement was super straightforward.

Once that was in the pen drive came back to life and worked brilliantly. It seems to use bumping into the first position to figure out where it’s at making it noisy and hilarious to use.

Download Geneva Drive

Once the carousel was working as expected I could tackle a bigger issue, pens. It turns out plotter pens aren’t nearly as common as one would hope. They are hard to find and really expensive so I decided to figure out a more maintainable solution. Luckily my plotter came with a couple of dried up pens and some not-so-dried ones so I could use them to figure out what I should look up for in a pen replacement. I first thought a ballpoint pen would be a good replacement, they are cheap, easy to find and replace, and some draw pretty nice lines.

I 3D printed a ballpoint pen holder shaped like a plotter pen. I got a couple of cheap ballpoint pens and stole their ink cartridges and went to town. It worked well but the vibrations seem to yield some shaggy lines. I decided to try something closer to the original next, I thought a felt tipped pen would probably be a better closer-to-the-real-deal option so I made an adapter for one of these Stabilo pens:

Download Stabilo Adapter

At the end I really liked how the ballpoint pen worked but I think I can still get a better line out of a finer Stabilo pen. Yet to test it so it'll have to wait for a later post.

HPIB

HPIB is an awesome little protocol that’s not very common outside of test equipment. The adapters are weird and expensive. If you have old test equipment lying around you might benefit from getting one along with a new plotter just cause. Don’t let that scare you away. I personally use an old prologix HPIB to Serial USB adapter and it works great. No regrets. Don't destroy an old plotter before giving this a go.

Testing OCLIF apps with Jest

Fri, 25 May 2018 00:00:00 +0000

A couple of weeks ago I spent sometime migrating Beau from commander to OCLIF. I really liked how every command in OCLIF is a stand alone class, it just felt tidier. Migrating the CLI was a breeze and I had the whole thing running in a couple of hours.

After I was done I decided I wanted to add some tests. I mostly cared about the output so all I wanted to do was snapshot it and let Jest warn me if it changed by accident. I mostly write tests to make rewriting things easier so snapshotting covers about 95% of my testing.

OCLIF recommends you use mocha and fancy-test for testing apps which is fine but since I was already using Jest I didn't want to add even more dependencies. The test helpers OCLIF provides seem nice, they mostly handle stdout and stderr, and http requests. These are handy but since I only cared about the output I figured I wouldn't need them.

OCLIF makes testing super easy. Every command is a class and every class has a static run method which you can use to, you guessed it, run the command. It takes an array with the arguments the command would normally receive. Given that, testing is made super easy.

Let's create a Test command that prints "test" and test that it does so.

Here's our test command:

const { Command } = require('@oclif/command');

class TestCommand extends Command {
	async run() {
		console.log('Test');
	}
}

module.exports = TestCommand;

If the CLI was called hello and we ran hello test we'd get 'test' back. Let's add a test to confirm we are getting the expected result back.

describe('Test Command', () => {
	it('should print Test', async () => {
		let result = await TestCommand.run([]);
		expect(result).toBe('Test');
	});
});

This is cool, but wouldn't work. Unfortunately the output is written to the STDOUT so that means that run doesn't actually return anything. Luckily for us Jest has spies built-in so all it takes is to add a spy to that call:

...
	it('should print Test', async () => {
		let spy = jest.spyOn(process.stdout, 'write');

		await TestCommand.run([]);
		expect(spy).toHaveBeenCalledWith('Test');
	});
...

To simplify this you can move the spy to a beforeEach.

describe('Test Command', () => {
	let result;

	beforeEach(() => {
		result = [];
		jest
			.spyOn(process.stdout, 'write')
			.mockImplementation(val =>
				result.push(val)
			);
	});

	afterEach(() => jest.restoreAllMocks());

	it('should print Test', async () => {
		await TestCommand.run([]);
		expect(result).toContain('Test')
	});
});

Whenever process.stdout.write is called you push the value into the result array. You can then check if the correct values are a part of the array. I'd just snapshot it and call it the day though.

Remember that the array that run expects is supposed to be the argv for that command. So they should be in the order they are expected and should all be strings.

Have fun testing your CLIs.

Hardware Debugging: Fixing My HP6236B

Wed, 14 Feb 2018 00:00:00 +0000

I recently added an HP6236B to my small but ever-growing vintage test equipment collection. It was in amazing condition when shipped but got smashed quite a bit on its way to me. The initial test wasn't promising:

{% video initial_test.mp4 %}

The insides looked really clean, and nothing seem out of place so I started going through the manual's debug steps. According to the manual a voltage reference should have been broken. So I started there. After replacing the reference the power supply behaved exactly the same. It behaved exactly the same even without the voltage reference in place so it had to be something else. I then proceeded to go through all the test points the manual had marked. All 15 of them seemed normal.

This took a long time and didn't really lead anywhere. I gave up and put the power supply away for a couple of days. The next time I got around to tinkering around with it I decided it was time to see what Jake, my oscilloscope thought about it.

Something was immediately obvious now, the power supply was oscillating quite a bit. Oddly enough, the -20v rail was actually working as intended but with it's range limited to what the 20v rail was stuck to. This was my first major clue. A quick check on the schematic showed that the likely culprit were the filtering capacitors near the output (C1 and C11).

I went ahead and removed them, replacing C1 with the closest capacitor I could find, a 220µF one. It was close enough. The power supply came to life and started working almost as intended.

{% video initial_working.mp4 %}

The next day I made a quick visit to a local electronic store and got the capacitors that needed replacement. A quick soldering job later and power supply was back to working as intended. I started fixing some of the broken pieces, like the meter holders. Those were not salvageable so I designed and 3d printed new ones.

Download Meter Holders

Once all the pieces were in, I resoldered the meters, placed everything back together and closed it up. I did some load testing and calibration using my Kikusui PLZ72w and it was good to go.

{% video final_working.mp4 %}

It found a nice place right next to my DC load.

Beau 0.4

Mon, 04 Dec 2017 00:00:00 +0000

Beau 0.4 is here! It includes some small goodies for ya'll. Mainly focusing on two things this time.

First, Beau now allows you to define requests in one line. If you pass a string as a configuration value it'll be taken as that request's alias so, you can get those small repetitive requests out of the way ASAP.

GET /users/:
  alias: get-users

GET /user/$get-users.body.0.userId:
  alias: get-user-0

Becomes:

    GET /users/: get-users
    GET /user/$get-users.body.0.userId: get-user-0

Much nicer.

The second change is the addition of a couple of extra flags to Beau's CLI: --clean-request and --clean-list . These print results for request and list without any formatting. This is mostly to make it easy to integrate Beau with other tools.

Speaking of other tools I am releasing a very-much in beta plugin for sublime .

The plugin allows you to make requests right from the Beau config file you are currently editing. There are several improvements I am currently working on which should be released soon. You can install it by using package control in sublime and searching for "Beau". Any feedback is welcome at the plugin's repository: https://github.com/Seich/sublime-beau/issues .

That's it for now, I am working towards releasing a more polished release before the year ends if I get the time. I have lots of new and nice things planned for Beau in the near future.

Writing Plugins for Beau

Thu, 06 Apr 2017 00:00:00 +0000

Currently Beau allows plugins to do two different transformations. You can change the request settings right before they are used to make the request and you can transform the response after the request is finished. This covers all use cases I've come up with so far. I'll go over the basics or writing a plugin in this post.

The Plugin

The first step is writing the plugin. A Beau plugin is a Javascript class that defines any of these methods: preRequest(requestSettings, originalRequest) and postResponse(response). They can also receive settings as part of their beau.ymlconfiguration. Here's the basic skeleton for a new plugin:

class MyPlugin {
  constructor(settings = {}) {
    this.settings = settings;
  }

  preRequest(req, orig) {}

  postResponse(res) {}
}

module.exports = MyPlugin;

All you are missing is a basic package.json to finish the module (just run npm init and go through the steps). Once that's done you can install your plugin globally and use it as part of your beau.yml like this:

Host: http://localhost:9000
   plugins:
       - MyPlugin:
           data:
               name: Sergio
           secret: ssh

GET /profile:
   alias: profile

GET /:
   skip_jwt: true
   alias: home

Our plugin will generate a jwt using the secret and data settings and send that along with every request. If skip_jwt: true is part of the request's configuration it won't attach the header. First I'll install jsonwebtoken:

npm install jsonwebtoken -S

Now all I need is to modify the preRequest so that it changes the headers and includes an Authorization with the bearer token:

preRequest(req, orig) {
    let token = jwt.sign(this.settings.data || {}, this.settings.secret);

    req.headers.Authorization = `Bearer ${token}`;
}

This will attach the authorization header to every request. We should also handle the case were certain requests don't need the header. In this case, I'll just check if the original request configuration has the field:

preRequest(req, orig) {
    if (orig.skip_jwt) {
        return;
    }

    let token = jwt.sign(this.settings.data || {}, this.settings.secret);

    req.headers.Authorization = `Bearer ${token}`;
}

That's it. Like I said before, it's pretty simple. This plugin as-is allows us for automatic token generation for certain requests. We can get as fancy or keep it as simple as we want. For now, I'll publish that plugin as beau-jwt and will make more improvements as I see fit.

Conclusion

Writing a plugin is pretty easy. Beau doesn't really do much so modifying requests is pretty straightforward and allows for nice, clean configuration files. Hopefully someone will think of cool things to do with these.

Hey Beau!

Mon, 13 Mar 2017 00:00:00 +0000

So I've been doing a lot of random experiments lately. One of them is Beau. Beau is at it's heart a configurable http client. It was built with JSON APIs in mind. I've used things like Paw or Postman in the past and they are good but they are lacking in some key areas. For example, I love how extensible Paw is but not being able to share project files with coworkers in Windows or Linux makes it so I can't use it with most of the teams I've worked with.

So to scratch that itch, I decided to build a simple http client for myself. I only had a couple of requirements. It should run on all platforms. It should be easy to setup and get started. Ideally the configuration is a plain text file so I can version control it and share it with others.

Having all of those things in mind I created Beau. Beau allows you to write a simple YAML file as configuration. It has a client written in Javascript which you can install using npm or yarn in a couple of seconds and it is completely cross-platform.

A simple beau configuration might look something like this:

POST /session:
    alias: session
    payload:
        username: seich
        password: password

With this simple config file, I can request a new session using the CLI:

beau -r session

That's all it takes. It also has a couple other tricks up it's sleeve though. You can also reference requests, so if that session request returns a token to verify other requests you can reference it like this:

GET /profile:
    alias: profile
    headers:
        authorization: Bearer $session.body.token

Beau understands $session is another request and it'll execute it first before attempting to do the profile request. The command doesn't change:

beau -r profile

Beau is still a work in progress but I feel like there is some potential to it. I'll be improving it's codebase and making it more extensible as time goes on. I am always open to feedback wether good or bad.

P.S. The incredibly talented, Angela Velasquez helped me out with a logo. Everyone knows a successful open source project begins with a great logo.

Making a Wooden iPhone Stand

Sat, 05 Dec 2015 00:00:00 +0000

I recently decided to tried my hand at building a physical thing. A stand for my iPhone. I decided I wanted to go through the whole process and see what its like. So I started with a mockup:

It’s a simple thing, a piece of wood with a slanted channel to put your phone in. Here’s how it turned out:

It looks ok, but there’s a couple of things I didn’t like with this first version. It held the phone just fine but it didn't look right or inspire enough confidence to actually trust it with my phone.

Time for a second version. There were three things I wanted to change. The width of the base, its angle and I wanted to add little anti-slip pads.

The width was straight forward, I measured my phone and added a centimetre. This way my phone would get to sit in the middle of the base.

For the angle, I wanted my phone to “face” me. The whole point of the base was to keep me from moving my neck too much to see my phone. This angle also defined the out cuts on the base because I wanted all the angles to match. After a bunch of test, I settled with 15º, which is rather comfortable.

Finally, I wanted to add anti-slipping pads to the base. I originally wanted to get some clear vinyl pads. I ended up settling for 1/2” cork pads though. The cork pads looked nicer and were thinner than their vinyl counterparts.

With everything figured out, I moved onto the mock up. I learned a couple of new techniques for sketchup so making it was a lot faster. Here’s what the mockup ended up looking like:

And here’s what it ended up looking like at then end:

All in all, I am pretty happy with the results. The base matches my original idea and fulfils my expectations. It’s not perfect, my wood working skills leave much to be desired. I made a couple of mistakes which I plan on correcting on my next build. Here’s the sketchup model in case you want to play with it or build your own: Download SketchUp Model

Postgres: Checking Queries in Real Time

Wed, 10 Jun 2015 00:00:00 +0000

Recently I came around a really simple way of checking what queries are going on in Postgres at any given time. Basically what you want to do is tell Postgres to log all queries and then you can tail the log file.

The first step is to tell Postgres to log all queries. To do that you go to your postgresql.conf file (mine’s here: ~/Library/Application Support/Postgres/var-9.4) and set the following options:

logging_collector = on
log_directory = ‘pg_log’
log_statement = ‘all’

Once you’ve configured these values Postgres will be dumping all queries into the current log file at pg_log. Now you can tail your log file and see what queries are happening in real time.

# Change this to your log’s folder.

tail -f ~/Library/Application\ Support/Postgres/var-9.4/pg_log/*

Also, it’s worth noting that Pg Admin has a log viewer you can use for this task. Just go to Tools > Server Status and you should get this window:

Expiring full-page caches in Rails 4 from the model

Sun, 08 Mar 2015 00:00:00 +0000

Recently I was struggling with Rail’s full-page cache. Ever since it was moved out of core I can’t seem to get sweepers to work correctly. After fiddling around for several hours I finally found the best solution possible (excluding sweepers, that is), invalidating the cache on the model’s callbacks. Here’s how it looks:

class Ad < ActiveRecord::Base
  after_save :clear_cache
  after_destroy :clear_cache

  def clear_cache
        ActionController::Base.expire_page('/ads.json')
  end
end

There’s a couple of gotchas:

You can’t use the nice syntax that’s available on controllers when you use expire_page, you have to pass the route to the file as a string. I couldn’t get it to work any other way (You could use url helpers to come up with this route, I suppose).
You have to make sure to expire any additional pages that might have changed base on the model’s relationship since this won’t handle it for you.

Other than these 2 things it’s pretty straight-forward and easy to use. I hope I don’t have to use full page caching again any time soon.

Angular.js One-way bindings

Fri, 26 Dec 2014 00:00:00 +0000

This week I discovered One-way bindings in Angular.

Angular's one-way bindings are a great way to improve your app's performance. Here's what the documentation says about them:

One-time expressions will stop recalculating once they are stable, which happens after the first digest if the expression result is a non-undefined value.
-Expressions, Angular Developer Guide

This means that by using them, once the value is not undefined, the bound value is removed from the watch list and thus isn't processed anymore during the digest cycle.

The syntax is pretty straight forward and very easy to use. All you have to do Once the value is set to the user's name, it won't be watched for changes anymore.

Great, huh?

You can use them as part of expressions and directives as well so, go crazy!

JustVector Social Font 1.5

Mon, 15 Sep 2014 00:00:00 +0000

Today I am releasing a new version of the JustVector social font.

3 years and 59k downloads later, the process of building icon fonts has changed radically. There are a bunch of new and better ways to get things done, I think the process has come a very long way in the past 3 years. Taking these new techniques and ideas into account, I've rebuilt the font from scratch.

I had a couple of objectives with this new version: First, I wanted to fit in all of the icons, when I first built the font this wasn't quite as trivial since I kept running out of glyphs I could replace but these days it's quite easy. You can use special glyphs and css to have as many icons as you want (I am not sure how many that is to be honest).

Second, I wanted to make the process of building a new font as automatic as possible. 3 years ago I had to manually export each glyph and add it to the font. It took fairly long and was painful to hint and size correctly.

The process is now automatic, I use font custom which allows me to automate the whole process. I can now build a new version in a couple of minutes. This means that I no longer will take 3 years between versions. I am aiming to keep it synced with the original set.

All in all I am really happy with the much-needed update. I'd like to thank Max Rice for helping me find the motivation to revisit the project and Alex Peattie for the wonderful icon set.

The project can now be found on Github: https://github.com/Seich/JustVector-Icons-Font

Here's a preview: Preview

Here's the download link: Download

Using TableViewControllers with Swift

Thu, 31 Jul 2014 00:00:00 +0000

Using a TableViewController with swift is easy. It takes only a couple of steps to create a table with custom cells. I started by creating a single view application.

Setting up the table and cell prototype

I started by deleting the default view. Then, I added a Table View Controller to the storyboard. Finally, I added two labels to the prototype cell. You can use the prototype cell to define how the cells for the table will look, so go ahead and tweak it to your convenience. Afterwards, I set the cell's identifier to 'cell'. I am creating a Twitter client so I only need to display the user and the Tweet.

(There's a preset style that's only a title and a subtitle which I could have used but, I want more control over it's appearance so I went this way)

Once I had my cell and table ready I just needed to do a couple more things. First, I created a class called TweetCell that inherits from UITableViewCell. I'll be using this class as a base for my custom cell. I set my class as the custom class and I added two outlets to refer to both of the labels. I named the big one "username" and the smaller one "body".

// TweetCell.swift
import UIKit

class TweetCell: UITableViewCell {
    @IBOutlet weak var username: UILabel!
    @IBOutlet weak var body: UILabel!
}

The ViewController

Now I only have to set up the view controller. We already have the basic view controller that came with the project so now we can just change it to do our bidding.

I start by creating a struct to contain my tweets. I added it before the ViewController's class definition.

struct Tweet {
    var username: String
    var body: String
}

After that, I changed the ViewController's parent class from UIViewController to UITableViewController and added the UITableViewDataSource protocol to it so my class looks like this:

class ViewController: UITableViewController, UITableViewDataSource {

}

Now I added an array with a couple of tweets:

var tweets: [Tweet] = [
    Tweet(username: "Seich", body: "Hello, World"),
    Tweet(username: "Albert", body: "How are you doing?")
]

Finally I just need to override a couple of methods. The first one is this method: tableView(tableView: UITableView!, cellForRowAtIndexPath indexPath: NSIndexPath!) -> UITableViewCell! which determines what cell should be drawn at the given index. Here's how mine looks:

override func tableView(
    tableView: UITableView!,
    cellForRowAtIndexPath indexPath: NSIndexPath!)
    -> UITableViewCell! {

    var cell = tableView.dequeueReusableCellWithIdentifier("cell") as TweetCell

    cell.username.text = tweets[indexPath.row].username
    cell.body.text = tweets[indexPath.row].body

    return cell
}

The second method is this one: override func tableView(tableView: UITableView!, numberOfRowsInSection section: Int) -> Int which should return the number of cells to draw, mine looks like this:

override func tableView(
    tableView: UITableView!,
    numberOfRowsInSection section: Int) -> Int {

    return tweets.count
}

And that's it. Here's my final code for the view controller:

import UIKit

struct Tweet {
    var username: String
    var body: String
}

class ViewController: UITableViewController, UITableViewDataSource {
    var tweets: [Tweet] = [
        Tweet(username: "Seich", body: "Hello, World"),
        Tweet(username: "Albert", body: "How are you doing?")
    ]

    override func tableView(
        tableView: UITableView!,
        numberOfRowsInSection section: Int)
        -> Int {

        return tweets.count
    }

    override func tableView(
        tableView: UITableView!,
        cellForRowAtIndexPath indexPath: NSIndexPath!)
        -> UITableViewCell! {

        var cell = tableView.dequeueReusableCellWithIdentifier("cell")
                        as TweetCell

        cell.username.text = tweets[indexPath.row].username
        cell.body.text = tweets[indexPath.row].body

        return cell
    }
}

And this is what my finalised app looks like:

It's not perfect but, it shouldn't take much to have it looking decent. I'll probably be revisiting this soon and finishing the whole Twitter application so stay tuned for more.

Resources

About Table Views in iOS Apps

Creating and Configuring a Table View

Storyboard Tutorial

Using MapKit with Swift

Tue, 29 Jul 2014 00:00:00 +0000

I've been using Swift a lot lately. I wanted to start making iOS apps a while ago but Objective-C is frankly, not worth the trouble. Swift on the other hand is actually fun to use and I love experiencing how the language evolves.

This weekend I started working on a small app I've wanted for a while but got stuck when it came to implementing a map with MapKit. I couldn't find documentation or code samples on how to do it anywhere. Turns out that it's not hard at all! It took very little fiddling around before I figured it out. Here's how I did it:

First, create a new project

I started with a Single View Application.

Then added a MKMapView to the view controller. After that, imported MapKit into the view controller's code. Finally, I added a new reference outlet to the map called map.

Setup your desired location

Now our map is ready to display a location. You'll need a couple of things to set it to the right place. First, you need a location in the form of a latitude and longitude and a CLLocationCoordinate2D.

var location = CLLocationCoordinate2D(
    latitude: 16.40,
    longitude: -86.34
)

Now, we need to tell the map what the area spanned by the region is. For this we use MKCoordinateSpanMake. Which, as you probably guessed, creates a MKCoordinateSpan.

var span = MKCoordinateSpanMake(1, 1)

After that we just need to define the region using these two things:

var region = MKCoordinateRegion(center: location, span: span)

And finally, we set this region to the map:

map.setRegion(region, animated: true)

Bonus: Adding an annotation

Adding an annotation is actually quite easy. We start by creating a new MKPointAnnotation instance. It'll display the default red pin which, might not be as pretty but it'll get the job done.

var annotation = MKPointAnnotation()

Afterwards, we can set it's coordinates, title and subtitle to place it wherever we want and to add the text it's callout will display when tapped.

annotation.coordinate = location
annotation.title = "Roatan"
annotation.subtitle = "Honduras"

Finally, we add it to the map like this:

map.addAnnotation(annotation)

Here's my final code:

import UIKit
import MapKit

class ViewController: UIViewController {

    @IBOutlet weak var map: MKMapView!
    override func viewDidLoad() {
        super.viewDidLoad()

        var location = CLLocationCoordinate2D(
            latitude: 16.40,
            longitude: -86.34
        )

        var span = MKCoordinateSpanMake(0.5, 0.5)
        var region = MKCoordinateRegion(center: location, span: span)

        map.setRegion(region, animated: true)

        var annotation = MKPointAnnotation()
        annotation.coordinate = location
        annotation.title = "Roatan"
        annotation.subtitle = "Honduras"

        map.addAnnotation(annotation)
    }
}

And here's what my app looks like:

Additional Resources

Location and Maps Programming Guide

Using mailgun to receive emails with Phabricator

Mon, 21 Apr 2014 00:00:00 +0000

Phabricator is awesome. We recently started trying it out for 2build.it (more like, this morning) and we all (Jorge and I) agreed it's awesome.

The setup is pretty straight-forward and only took like 20 minutes. The one thing I got stuck on though, was the emailing. It took me forever to realise that you're supposed to restart the emailing daemons for the configuration to stick. It is my fault but, this wasn't mentioned anywhere in the documentation. Btw, here's how you do it:

./bin/phd restart

Right from the Phabricator folder.

Moving on.

Now that sending was settled there was one huge thing that wasn't working. Inbound emails. I tried working with postfix but, it was too painful and after a couple of hours I didn't get anywhere. The documentation talks about using sendgrid to achieve it but, I already have a mailgun account and according to the configuration Phabricator should be able to use mailgun as well. So, I had to do some research.

It turns out you can use mailgun to do this but, there's zero documentation about it. Luckily, I found the commit where this functionality was added and figured it out on my own. Here's the three step guide to getting it done.

Step 1

You have to configure you domain's MX records to point to mailgun. You can find more information here.

Step 2

Go to your setup your mailgun configuration by going to the mailgun config (http://example.com/config/group/mailgun/) and adding your API key and your domain.

Step 3

We're almost done. Now, all we have to do is tell mailgun to send received emails your way. To do this all you have to do is create a new route (go here: https://mailgun.com/cp/routes). Give it a priority and a filter that'll work for you (I used 0 as a priority and catch_all() for a filter). For the action, forward it to your Phabricator installation with the endpoint '/mail/mailgun'. Like this: forward("http://example.com/mail/mailgun/").

That is all. Your emails should now allow you to interact with Phabricator.

Sketch Presets for Desktop Screens

Sat, 05 Apr 2014 00:00:00 +0000

I've been using Sketch to create mock ups for clients lately. So far so good. Today, I wanted to create art board presets for common desktop screen sizes. It took me a little while to find out how to organize them into folders. Luckily for me, Jan Drewniak has a blog post detailing how to do just this.

Sketch actually keeps these artboard presets in an artboards.sketchpreset file located somewhere on your mac. If you're using the App Store version of Sketch, then the file is located here:

~/Library/Containers/com.bohemiancoding.sketch/Data/Library/Application  Support/sketch/

Here's my presets:

<dict>
	<key>name</key>
	<string>Desktop</string>
	<key>presets</key>
	<array>
		<string>Retina Displays</string>
		<dict>
			<key>width</key>
			<integer>2880</integer>
			<key>name</key>
			<string>15-inch</string>
			<key>height</key>
			<integer>1800</integer>
		</dict>
		<dict>
			<key>width</key>
			<integer>2560</integer>
			<key>name</key>
			<string>13-inch</string>
			<key>height</key>
			<integer>1600</integer>
		</dict>
		<string>4:3</string>
		<dict>
			<key>width</key>
			<integer>1600</integer>
			<key>name</key>
			<string>Large 4:3</string>
			<key>height</key>
			<integer>1200</integer>
		</dict>
		<dict>
			<key>width</key>
			<integer>1280</integer>
			<key>name</key>
			<string>Medium 4:3</string>
			<key>height</key>
			<integer>1024</integer>
		</dict>
		<string>16:9</string>
		<dict>
			<key>width</key>
			<integer>1920</integer>
			<key>name</key>
			<string>Large 16:9</string>
			<key>height</key>
			<integer>1080</integer>
		</dict>
		<dict>
			<key>width</key>
			<integer>1600</integer>
			<key>name</key>
			<string>Medium 16:9</string>
			<key>height</key>
			<integer>900</integer>
		</dict>
		<dict>
			<key>width</key>
			<integer>1366</integer>
			<key>name</key>
			<string>Small 16:9</string>
			<key>height</key>
			<integer>768</integer>
		</dict>
	</array>
</dict>

I have categories for: Retina, 16:9 and 5:4 displays. Only listing the ones I've used recently and might re-use in the future.

Scroll Spying for Fun and Profit

Fri, 04 Apr 2014 00:00:00 +0000

I recently redesigned by blog. One of the new features is a menu that highlights the current post you are on.

To achieve this, I used Zepto and a small amount of math. (The great thing about Zepto is that it's fully compatible with jQuery, so this should work with jQuery out of the box as well).

Let's jump right in, here's how I do it:

$(window).on('scroll', function() {
    var scrollTop = $(window).scrollTop();
    var article_offsets = $('article').map(function(i) {
        if (i === 0) {
            return 0;
        }

        return $(this).offset().top
    });

    for (var i = 0; i &lt; article_offsets.length; i++) {
        if (article_offsets[i] - 40 &lt;= scrollTop) {
            $('.sidebar li')
                .removeClass('active')
                .eq(i)
                .addClass('active');
        }
    };
});

The first line of interest:

var article_offsets = $('article').map(function(i) {
    if (i === 0) {
        return 0;
    }

    return $(this).offset().top
});

Grabs all the vertical positions (from now on, offsets) of the elements I want to match and makes an array out of them. In this case, I am targeting the articles since, those represent my blog posts. I make sure to mark the first one as 0, so it works when the page first loads at scroll position 0.

After doing this, we iterate over all our scroll positions, if we find ourselves scrolled after one of the offsets we run this:

    $('.sidebar li').removeClass('active').eq(i).addClass('active');

This looks for the Ith element on the sidebar's list and we add a class of 'active' to it, highlighting it and doing whatever we want with it.

There's a couple of other (bigger, more complex and probably more robust) solutions to this problem out there. Here's a couple I came through:

I am pretty sure they are plenty of others but, for my evil purposes, my tiny script fulfills my needs.

2014

Sun, 26 Jan 2014 00:00:00 +0000

I know I am a little late doing the whole goal setting and year review thing but, this year I wanted to try something different. I wanted to see how January turned out before setting goals.

This year will be pretty exciting. I'll continue working as a consultant as I have for the past couple of years. Invivo Link is doing some pretty cool things and I am happy to help them continue building their product. Over at 2build.it, we're working with some pretty cool clients, building a variety of pretty cool things, hopefully the company will grow a whole lot this year.

I tend to decide what I'd like to learn more during the year. This year's language is going to be Rust. I've fallen in love with it on the past couple of months and I'd like to start building more things with it as the language becomes more stable. Apart from Rust, this year I'll be working with Objective-C a whole lot more. I don't fully like the language yet but, I feel like it can be a valuable tool in my toolset. I'll also continue working on bundle.js as it has become essential to a client's project which, turns out to be a great way to further develop it.

I'd also like to write more, I didn't get to do it as much as I wanted to last year so I've adjusted my goal, I want to write at least, 12 posts during the year. I think this is a reasonable quantity and I hope I'll be able to reach it. I'd also like to read more, my goal for that is at least 52 books. I am up to speed so far so I am pretty sure I can reach that.

Finally, I'd like a travel more this year, I am getting a little tired and burned out and I feel that it would help me to travel and see things.

I achieved most of the important goals from last year so, I am looking forward to a great year.

You've Gotta Love Deferreds

Tue, 12 Nov 2013 00:00:00 +0000

Deferreds are awesome. They are such a simple solution to a really hard problem. Keeping things async in javascript can get hard sometimes. I recently wrote some code that looked like this:

var settings = {
	template: 'todo.hbs',
	data_src: '/todo.json'
};

var template;

var getting_template;
var getting_data;

if ('template' in settings) {
	getting_template = $.get(settings.template);

	getting_template.done(function(t) {
		template = t;
	});

	if ('data_src' in settings) {
		getting_data = $.get(settings.data_src);
		getting_data.done(function(data) {
			render(template, data);
		});
	} else {
		getting_template.done(function(t) {
			render(t, {});
		});
	}
} else {
	render('', {});
}

(I am pretty sure it wasn't necessarily as ugly but, I don't quite remember.)

As you can see, even though I am making use of the deferreds returned by the $.get function. This could be cleaned up a lot by using even more deferreds.

Here's more or less what the code ended up looking like:

var settings = {
	template: 'todo.hbs',
	data_src: '/todo.json'
};

var templateDefer = new $.Deferred();
var dataDefer; = new $.Deferred();

if ('template' in settings) {
	$.get(settings.template, function(template) {
		templateDefer.resolve(template);
	});
} else {
	templateDefer.resolve('');
}

if ('data_src' in settings) {
	$.get(settings.data_src, function(data) {
		dataDefer.resolve(data);
	});
} else {
	dataDefer.resolve({});
}

$.when(templateDefer, dataDefer).done(function(template, data) {
	render(template, data);
});

I basically kept everything async using stand-alone deferreds. Turns out jQuery has really handy deferred util built-in, all you have to do is create a new one (new $.Deferred()). The other handy thing built into jQuery is the $.when method, you can basically pass in a bunch of deferreds and you'd get a single one that responds when all of them respond in a particular way. That way, you can basically say, when they are all done, do this.

I shall try experimenting a lot more with these tools, they are pretty easy to use and pack in, so much functionality.

A Semantic Grid with Zurb's Foundation

Fri, 11 Oct 2013 00:00:00 +0000

Foundation is great. I love it, I use it whenever I am building anything big. My favorite feature and the one key thing that has kept me as a fan for so long has been it's grid. I feel like it's the best grid we have available so far.

My favorite thing about that grid you ask? Apart from just being plainly awesome? Well, it's the semantic option. I know you probably haven't put much thought into it but, don't you feel like grid classes tend to clutter the html? I do, and I hate it. Enter the mixins!

The Mixins

To alleviate this issue, Foundation gives us a set of really handy mixins which allow you to build a really semantic grid. It's as flexible as it can be and just as easy to use (I'd argue that if you're familiar with sass, it's just plainly easier and requires less typing).

To demonstrate, I'll be showing you how to build this:

As you can see, it's a very simple layout. It'd be really easy to build it using the grid classes, it's even easier to build it using the mixins. Here's my markup:

<article>
  <header>
    <h1>Hello World</h1>
  </header>

  <div>
    <img src="http://lorempixel.com/300/400/" />
  </div>

  <p>
    Lorem ipsum dolor sit amet, consectetur adipisicing elit. Exercitationem,
    voluptatibus, deserunt, facilis iusto sapiente impedit praesentium
    laudantium minus voluptas numquam incidunt corrupti sequi laboriosam magnam
    officia perspiciatis quas quam quaerat!
  </p>
</article>

As you can see it's pretty standard except the elements lack classes and such. Much prettier than usual in my opinion.

Here's my CSS:

@import "foundation/components/global";
@import "foundation/components/grid";

$total-columns: 5;
article {
  @include grid-row;

  header {
    @include grid-row;

    h1 {
      @include grid-column(5);
    }
  }

  div {
    @include grid-column(2);
  }

  p {
    @include grid-column(3);
  }
}

Awesome, right? Here I am the grid-column and grid-row mixins to create the grid directly in my CSS. All you have to do is import the mixins. There are also many options to customize the grid, in this case I changed the number of columns from the default 12, to 5. I am also nesting rows, all from the CSS, leaving the HTML to describe it's thing.

Side note:

I hadn't notice that in this case you're putting a row directly inside another row, this isn't ideal as it can cause alignment issues. Ideally you'd have the header inside full width column to prevent this. I don't think it matters in this example but in anything more complex it might be a pain to debug.

Thanks to @smiley for the heads up!

Conclusion

I feel like this technique is a great way to simplify the HTML and further decouple it from it's CSS. They've been several similar things before such as the Semantic Grid which experimented with this idea. Foundation has a much more powerful grid so, it being able to work this way as well makes me really happy.

Connecting To A Remote RethinkDB Admin Interface

Sat, 20 Jul 2013 00:00:00 +0000

So here's a really quick tip, when you run a RethinkDB instance on a remote server you cannot access the admin interface, by default, from the outside of the server. The easiest way to get around this without allowing anyone to access it is to use a ssh tunnel like this:

$ ssh -L 8080:localhost:8080 &lt;host&gt;

This basically uses ssh to map the port 8080 of your machine (127.0.0.1 or localhost) to the port 8080 on the host machine. Don't forget to replace "<host>" with your actual remote host.

With this, you can now connect to the admin interface by going to localhost:8080 on your machine. Note that this will only work as long as the ssh connection remains open.

Additional Resources

A Javascript Array of Countries

Fri, 28 Jun 2013 00:00:00 +0000

Here's the Array:

var countries = ["Afghanistan", "Åland Islands", "Albania", "Algeria", "American Samoa", "Andorra", "Angola", "Anguilla", "Antarctica", "Antigua And Barbuda", "Argentina", "Armenia", "Aruba", "Australia", "Austria", "Azerbaijan", "Bahamas", "Bahrain", "Bangladesh", "Barbados", "Belarus", "Belgium", "Belize", "Benin", "Bermuda", "Bhutan", "Bolivia, Plurinational State Of", "Bonaire, Sint Eustatius And Saba", "Bosnia And Herzegovina", "Botswana", "Bouvet Island", "Brazil", "British Indian Ocean Territory", "Brunei Darussalam", "Bulgaria", "Burkina Faso", "Burundi", "Cambodia", "Cameroon", "Canada", "Cape Verde", "Cayman Islands", "Central African Republic", "Chad", "Chile", "China", "Christmas Island", "Cocos (keeling) Islands", "Colombia", "Comoros", "Congo", "Congo, The Democratic Republic Of The", "Cook Islands", "Costa Rica", "Côte D'ivoire", "Croatia", "Cuba", "Curaçao", "Cyprus", "Czech Republic", "Denmark", "Djibouti", "Dominica", "Dominican Republic", "Ecuador", "Egypt", "El Salvador", "Equatorial Guinea", "Eritrea", "Estonia", "Ethiopia", "Falkland Islands (malvinas)", "Faroe Islands", "Fiji", "Finland", "France", "French Guiana", "French Polynesia", "French Southern Territories", "Gabon", "Gambia", "Georgia", "Germany", "Ghana", "Gibraltar", "Greece", "Greenland", "Grenada", "Guadeloupe", "Guam", "Guatemala", "Guernsey", "Guinea", "Guinea-bissau", "Guyana", "Haiti", "Heard Island And Mcdonald Islands", "Holy See (vatican City State)", "Honduras", "Hong Kong", "Hungary", "Iceland", "India", "Indonesia", "Iran, Islamic Republic Of", "Iraq", "Ireland", "Isle Of Man", "Israel", "Italy", "Jamaica", "Japan", "Jersey", "Jordan", "Kazakhstan", "Kenya", "Kiribati", "Korea, Democratic People's Republic Of", "Korea, Republic Of", "Kuwait", "Kyrgyzstan", "Lao People's Democratic Republic", "Latvia", "Lebanon", "Lesotho", "Liberia", "Libya", "Liechtenstein", "Lithuania", "Luxembourg", "Macao", "Macedonia, The Former Yugoslav Republic Of", "Madagascar", "Malawi", "Malaysia", "Maldives", "Mali", "Malta", "Marshall Islands", "Martinique", "Mauritania", "Mauritius", "Mayotte", "Mexico", "Micronesia, Federated States Of", "Moldova, Republic Of", "Monaco", "Mongolia", "Montenegro", "Montserrat", "Morocco", "Mozambique", "Myanmar", "Namibia", "Nauru", "Nepal", "Netherlands", "New Caledonia", "New Zealand", "Nicaragua", "Niger", "Nigeria", "Niue", "Norfolk Island", "Northern Mariana Islands", "Norway", "Oman", "Pakistan", "Palau", "Palestine, State Of", "Panama", "Papua New Guinea", "Paraguay", "Peru", "Philippines", "Pitcairn", "Poland", "Portugal", "Puerto Rico", "Qatar", "Réunion", "Romania", "Russian Federation", "Rwanda", "Saint Barthélemy", "Saint Helena, Ascension And Tristan Da Cunha", "Saint Kitts And Nevis", "Saint Lucia", "Saint Martin (french Part)", "Saint Pierre And Miquelon", "Saint Vincent And The Grenadines", "Samoa", "San Marino", "Sao Tome And Principe", "Saudi Arabia", "Senegal", "Serbia", "Seychelles", "Sierra Leone", "Singapore", "Sint Maarten (dutch Part)", "Slovakia", "Slovenia", "Solomon Islands", "Somalia", "South Africa", "South Georgia And The South Sandwich Islands", "South Sudan", "Spain", "Sri Lanka", "Sudan", "Suriname", "Svalbard And Jan Mayen", "Swaziland", "Sweden", "Switzerland", "Syrian Arab Republic", "Taiwan, Province Of China", "Tajikistan", "Tanzania, United Republic Of", "Thailand", "Timor-leste", "Togo", "Tokelau", "Tonga", "Trinidad And Tobago", "Tunisia", "Turkey", "Turkmenistan", "Turks And Caicos Islands", "Tuvalu", "Uganda", "Ukraine", "United Arab Emirates", "United Kingdom", "United States", "United States Minor Outlying Islands", "Uruguay", "Uzbekistan", "Vanuatu", "Venezuela, Bolivarian Republic Of", "Viet Nam", "Virgin Islands, British", "Virgin Islands, U.s.", "Wallis And Futuna", "Western Sahara", "Yemen", "Zambia", "Zimbabwe"];

It follows the ISO 3166-1-alpha-2 standard. You don't need an array with countries everyday but, when you do, getting it is a pain. I am putting it here so my future self doesn't have to look for it again.

Here's how I got the list:

$ = jQuery;
String.prototype.capitalize = function() {
    return this.replace(/(?:^|\s)\S/g, function(a) { return a.toUpperCase(); });
};

var countries = "var countries = [";

$('tbody').find("tr").each(function(i) {
    var country = $(this).find('td').last().text();
    country = country.toLowerCase().capitalize();
    if (i === 0) {
        countries = countries + "\"" + country + "\"";
    } else {
        countries = countries + ", " + "\"" + country + "\"";
    }
});

countries = countries + "];";
console.log(countries);

If you run that script in here, you'll get a nice string printed on the console which you can copy and paste and use as an array. Quick 5-minute hack FTW.

Design For Your Mother

Wed, 20 Feb 2013 00:00:00 +0000

I am not really that much of a designer. I've worked as one but, I'd never describe me as a "web designer". I am simply not that good with designing things. I do have good taste though and, can tell good design apart from bad design. I also have a solid knowledge of design and UX in general which is enough for my day to day activities. At the end of the day, the only thing I know about design, is that good design should be easy to use. How easy? Well, my mother should be able to use it without having to ask for my help. I obviously know that not all mothers need help when it comes to computers (to be honest, mine can handle herself just fine without me) but, the point is, your design should be easy to use first, and nice-looking second. My mother should be able to use your application without having to think too hard. Users are lazy and they like being able to do stuff without thinking. Try as hard as possible to make it easy for them to not think.

I really like how simple interfaces are kind of trending these days. It really is great, simple design helps get the message through and allows users to interact more freely with the web. I've always gone with "minimalist" designs mostly because, it's easier to get right when you are not that great of a designer and also because it's easier for the user to interact with any application if they are only presented what they need at any given moment.

Are you sure you need that extra button? Would anyone apart from power users benefit from having this on the interface? Will this get the message through? Can we hide all of this complexity so they can find it if they need it but wont get in the way if they don't?

Think about, simple design is awesome, don't make me think and don't make my mother ask me for help.

Organizing a Javascript MVC Project

Fri, 01 Feb 2013 00:00:00 +0000

I've been doing a lot of javascript development lately. My work as a frontend developer for Laureate has forced me to seriously how we organize the frontend code so it's easy to work with and hard for us to make mistakes.

As it stands, I developed the frontend structure of the project (I am not sure if I am allowed to share it) we are currently building. We use can.js as our MVC framework of choice and use grunt to stick all the files together and check them for errors. All this, in the effort to be as agile as possible.

Coming up with a structure was pretty hard but, I think out current structure works pretty well. Our folder structure looks somewhat like this:


    -Project
        -Application
            -Assets
                -Images
                -Css
            -Javascript
                -Models
                -Views
                -Controllers
                -libs
                    -Plugins
                    jQuery.js
                app.js
        -dist
            -app.min.js
            -libs.min.js
        index.html
        grunt.js

It might look complicated, but it's actually fairly simple. I'll try to break it down, show how we use grunt to generate app.min.js and libs.min.js and how they all generally fit in together.

Folder & File Structure

The folder structure is fairly straight forward. Inside the Application folder we hold all of our important files. The only important folder here is the one called Javascript, this is where we store all of our Models, Views and Controllers which at the end, make up our entire application. You can have more folders inside of these and grunt can still fetch all of their contents to build up the final files.

Here's what a raw controller inside this folder looks like:

;(function(namespace, undefined) {
  'use strict';
  var ControllerName = can.Control({
    'init': function(element, options) {
      var self = this;
      // Actual code stuff
    }
  });

  namespace.Controllers = namespace.Controllers || {};
  namespace.Controllers.ControllerName = ControllerName;
}(this));

As you can see the only thing we are doing which isn't immediately obvious is to add the controller to the namespace's Controller object. In our case, this is the window object. This is great because we don't populate the global namespace with every single controller and because we can access any controller like this: Controllers.ControllerName. We do the same for models:

;(function(namespace, undefined) {
  'use strict';
  var ModelName = can.Model({
    findOne: 'GET /getModel'
  }, {});

  namespace.Models = namespace.Models || {};
  namespace.Models.ModelName = ModelName;
}(this));

In our case, views are simple .ejs files so there's nothing special about how we define them, although you could compile them to javascript and tell grunt to minimize it along with the whole code which would probably faster to load in the long run.

Now that we have our files in place, let's move into using grunt to generate the finalized application files will be using in production.

Grunt

Here's what an example grunt file would look like:

module.exports = function(grunt) {
  grunt.initConfig({
    concat: {
      app: {
        src: [
          'Application/Assets/Javascript/controllers/**/*.js', 
          'Application/Assets/Javascript/models/**/*.js', 
          'Application/Assets/Javascript/app.js'
        ],
        dest: 'dist/app.js'
      },
      libs: {
        src: [
          'Application/Assets/Javascript/libs/jQuery.js', 
          'Application/Assets/Javascript/libs/plugins/**/*.js'
        ],
        dest: 'dist/libs.js'
      }
    },
    min: {
      app: {
        src: ['&lt;config:concat.app.dest&gt;'],
        dest: 'dist/app.min.js'
      },
      libs: {
        src: ['&lt;config:concat.libs.dest&gt;'],
        dest: 'dist/libs.min.js'
      }
    }
  });

  grunt.registerTask('default', 'concat min');
};

This particular grunt file doesn't do much. When you call the grunt command it'll concatenate all of the files in the javascript folder (except for the views) and all of the libraries into two files, libs.js and app.js (and their minimized equivalents). These can be included directly into the html to have the final application files.

That's it

I hope this is somewhat useful to someone, it actually took me a while to find something I was comfortable with so try experimenting with variations and let me know about it. This is how I currently do things but, I am sure there are places where I can improve the design. This mostly revolves around keeping the global namespace as clean as possible and making files that can be easily concatenated to save space when you have to finalized copy which, I think are generally good goals.

Surviving a Hackathon

Sun, 27 Jan 2013 00:00:00 +0000

Last week I finally went to my first hackathon and it was an awesome experience! Here are a couple of things everyone should watch out for when they go to a hackathon for the first time.

What you'll need

Okay, first things first, hackathons are apparently seriously social events. If you are more of an introvert it might be good for a you to bring along a friend, it'll make your time there easier. I brought a friend along and it made the whole thing a lot more fun.

There are other things which are kind of pretty obvious stuff such as: a laptop if you are going to be doing some coding and headphones if you need to block the world to get things done (Although, like I said, these are social events so try not to get too lost in your code and enjoy the company). Cameras are also great, your phone will probably do if you don't feel like carrying additional stuff.

Finally, another thing, which I didn't actually bring with me on purpose are business cards. Even though I was just hanging around and doing stuff a bunch of people approached me with project ideas and business offers. So if you are looking for a job that might also be a good place to look for one. I almost ran out of them during the weekend!

Hackathoning!

Hackathoning is the only reason you are awake at 9am on a Saturday so it's important to keep it in mind. If there's a theme which was set beforehand, I'd recommend you go ahead and think of ideas so you can discuss them with your teammates. If there's no theme then thinking of ideas is still a good idea.

The first thing you'll need to do if find a team, if you wanna work on your own (which I don't think is a good idea) or brought along your own team, you are done with this and can go ahead and start brainstorming. If you didn't and have to find a team look for people who can complement your skill-set. Let them know the kind of thing you wanna work on and what you can do or want to do.

Once you are in a team, you are ready to make a plan. Make sure the plan is feasible, you only have a limited amount of time after all so, if it's too complicated you might have enough time to finish, you can always continue working on the project after the hackathon so, keep it achievable.

At this point, you are ready to start coding! So go out there and code! Try not to get too distracted or too focused, try to socialize and get stuff done at the same time. If you can balance these things you'll survive this hackathon successfully.

Networking

This is, what I found to be, the most important aspect of the hackathon. You cannot only socialize but, you can also make new friends, business partners, clients and jobs. So try to talk with everyone, it's always nice to find new project to work on.

Closing up

Hackathoning is fun! You can network, socialize, code and have a general good weekend so, if you haven't gone to one, I highly recommend you do.

Creating a model while skipping validations in Rails

Fri, 18 Jan 2013 00:00:00 +0000

This certainly needs some background. I've been working on a random rails project for the better part of the last couple of months. This project uses devise to manage logins and user registration. Recently registrations started failing in the site which, didn't make sense since they had been working fine for a while. So I started digging into it.

According to rails the user url didn't exist so it couldn't redirect back to the homepage after a successful registration. After much digging around the stack trace I finally figured it out. Turns out that when a devise user is created a user profile is created along with it(this is specific to my application). But, that's normal, what started causing problems was that the profile had a validation for a unique username, which blank, wasn't (since the username is only supposed to be set after registration).

So, what's the solution? Easy, all I had to make sure of was that when I created the user profile I disabled validations for the user profile model. Here's how I did it:

user.rb

class User < ActiveRecord::Base
  after_create :create_profile
  has_one :user_profile, :dependent => :destroy

  def create_profile
    self.build_user_profile
    self.user_profile.save(:validate => false)
  end
end

user_profile.rb

``` ruby class UserProfile < ActiveRecord::Base belongs_to :user

attr_accessible :username validates_uniqueness_of :username end

<p>This bug took around 2 hours to fix, I guess not all errors in rails are as descriptive and easy to follow as they should. But at the end of it all, it's my fault for not paying attention.</p>

Hiding elements when they lose focus with jQuery

Thu, 10 Jan 2013 00:00:00 +0000

There's no standard way of hiding an element when it loses it's 'focus' using Javascript or jQuery. Here's how I've been doing it lately:

$(".userMenu a").on("click", function() {
  $(".loginForm").stop().fadeToggle(100, "linear", function() {
    $("body").on("click.loginform", function(e) {
      if($(e.target).parents('.loginForm, .userMenu').length === 0) {
        $('.loginForm').fadeToggle(100);
        $("body").off(".loginform");
      }
    });
  });
});

As you can see, it's fairly straight forward. Basically, when you click the button that shows the element, it toggle's the target element's visibility. When it is visible, an event is bound to the body, so whenever it is clicked, it checks if the click was on the element or if it wasn't. If it wasn't on the element, we hide the element and remove the event from the body. If it was on the element, nothing happens.

So far, I like this method a lot, mostly because it is simple and gets things done quickly.

New Year's Resolutions 2013

Sat, 05 Jan 2013 00:00:00 +0000

I'll begin by saying this will be a great year. I have a lot of objectives and things I want to accomplish. Some will be the culmination of past efforts and some are new things I want to get done. My first resolution is to blog more. I've been saying I'll start blogging more often for a couple of years now and I keep failing. I'll change that and thus, my first objective is to write 52 posts this year, one per week. Since I don't alway have too much time I am not requiring myself to publish every week as long as I get it done eventually. I also will write small random stuff most of the time since, that's how I roll.

My second resolution is to start a company. I am currently participating in contest sponsored by my university. If I were to win I'd be established as a company in a couple of months. Even if this doesn't work out, I'll get started on my own.

There's more development and learning to be done in 2013 as well. I'd say 2012 was the year of Javascript. Working as a front-end developer forces you to learn a lot of Javascript so it's been a great experience. This year I'll further evolve as a Javascript developer but, I also want to master ruby. I just love that language and even though I've worked with it a lot I don't feel as confident in my skills. It still feels too magical for my taste. I'll be working a lot more on personal projects (and blogging about it as well). I already have one android app in mind and a couple of gems to make my life easier.

In the end I hope this year is as great as the last one and I learn, develop and grow a lot more.

A little House Keeping

Mon, 15 Oct 2012 00:00:00 +0000

Hello world! It's that time of the year again where I am forced to do a little house keeping. My blog has been incredibly hideous for a couple of months now and I finally got the opportunity to do somethign about it.

I've redesigned it and, since apparently I am terrible at designing stuff I went with a very minimalistic approach. So my new blog is easier on the eyes, simpler and, will now become my homepage.

There isn't much to it, I'll be moving a couple of my old posts over to the new blog and I'll do my best to start blogging at least twice a week. It's actually harder said than done...

The blog runs on Jekyll now, mostly because it is hacker friendlier than tumblr and I love the idea of version controlled blogging. I will also remove comments from the blog since they distract me quite often. If anyone feels the need to tell me anything they are free to email me.

Random stuff has been happening lately

Mon, 30 Jul 2012 00:00:00 +0000

Ok, so what’s up dear readers? I’ve been up to a bunch of random things lately so I’ll write them down just so I know what I was up to at this time of year later on.

First of all, I got a new job. I guess I’ll stop freelancing for the time being and focus on my new (awesome) job. Starting 2 weeks ago, I’ll be working as a “Graphic Developer” for the HNSC, which, is a fancy way to say I am a front end developer. I get to play with lots of javascript and are surrounded by amazing people I didn’t even know about! So yes, lots of new people and cool projects which might actually inspire me to build new and cool things on my own.

Second, I’ve been getting into javascript a lot lately. Like a serious lot. Even before my new job, I started reading and developing almost exclusively in javascript. I feel like I’ve been improving a lot constantly these past couple of weeks and am really happy about it. I’ve read and develop a lot, trying to get my vanilla javascript up to speed and I’ve been working with a lot of jQuery at work as well so that’s that. Today I built a jQuery plugin which, doesn’t do much but it’s exactly what I needed I’ll talk more about it once I figure out if I can release it as open source. That’ll hopefully happen soon.

Third, since I recently started working as a frontend developer, which is almost completely opposite from my old freelancing stuff where I focused on the backend, I decided to start a pears repository of stuff I use often just to keep mental notes and to share with you guys. You can find them here: http://patterns.martianwabbit.com/

And I guess that’s basically it, school is as hard as usual, I am still pretty dumb and I get paid to program for 4 hours on a daily basis, I am doing pretty good these days.

Playing with Arduinos, RFID cards and Ruby

Mon, 28 May 2012 00:00:00 +0000

This is probably one of the coolest things I’ve worked on in a while and I also happen to have time to write a blog post about it, which is great, I love working with hardware from time to time.

It definitely isn’t the biggest programming challenge I’ve had but it is pretty cool and I am certain others might find it interesting. Yesterday, I was helping my sister with a project she was doing for her professional practice (she is a very weird kind of engineer who specializes in automation stuff, she is great at it), she was basically asked to open a door with a RFID card.

To do this, she had a very simple plan, we plug an RFID card reader into a computer which then sends the RFID over to a computer and then tells an Arduino to open the door. She used this RFID Reader (RFID Reader ID-20 (125 kHz)) which is pretty inexpensive and easy to plug into a computer.

This was fairly straight-forward for her, she mounted the RFID card reader into a board and plugged it into her computer along with her Arduino. She then fired up putty and set it to read the RFID card reader’s serial port (COM 9, in our case) which showed us the RFID of any card we read.

This is where I came in, she isn’t that much of a coder so she asked me for help on how to finish the connection between the RFID reader and the Arduino. This is what I came up in Ruby:

require "serialport"

@rfid_reader = SerialPort.new 4 # Open COM 5
@arduino = SerialPort.new 2 # Open COM 3

@valid_ids = ['4500B8D690BB']

def readRFID
  rfid = @rfid_reader.gets
  rfid.strip!
  rfid.slice(0) if rfid.size == 13
  rfid.slice(0..1) if rfid.size == 14

  return rfid
end

puts "Waiting for input..."
loop do
  rfid = readRFID # Get the RFID of the current card
  if !rfid.empty?
    if @valid_ids.include? rfid # is the ID valid?
      puts "RFID: " + rfid

      # Send the arduino a signal so it opens the lock.
      @arduino.write "1"
    else
      puts "Invalid RFID: " + rfid
      @arduino.write ""
    end
  end
  @arduino.write ""
end

I was actually going to use C++ but I prefer using Ruby whenever possible. This is a very simple code snippet but it did the job wonderfully. The code is documented and very simple, I used the fabulous ruby-serialport gem which, took care of all the heavy lifting for me.

This was only half of the work though, we also needed to get the Arduino to wait for an input coming from the computer and do something whenever a value is received, in this case we wanted to open a lock so that’s what we’ll do.

void setup () {
  Serial.begin (9600);
  pinMode (12, OUTPUT);
}

void serialEvent() {
  digitalWrite(12, HIGH);
  delay(2000);
  digitalWrite(12, LOW);
  Serial.read();
}

The code is even simpler than the Ruby one running on the computer. All it does is wait for there to be an input and then activate a pin when there is one. Ruby is pretty much doing all of the actual processing and telling the Arduino when and what to do.

Finally, all we had to do was fire up the terminal and run our Ruby program which would then wait for there to be an input form the RFID reader. Whenever there’s an input it’ll log it and then, if the RFID is registered, tell the Arduino to open the lock. And that’s basically it, it made a nice problem to solve for fun.

Mew: Windows Markdown Editor

Sun, 04 Dec 2011 00:00:00 +0000

So for some mysterious and unknown reason I am selling a windows markdown editor. It’s called Mew, and you can get it for $8 at codecanyon.

Mew, as you’d expect from a markdown editor is pretty simple. I built it for myself using C# and the awesome .net framework. I was basically forced into building it since, there wasn’t a nice markdown editor for windows available. I worked on polishing it for a couple of weeks and now here the end result:

It has built-in snippets, color themes and a live-preview, among other things.

I am fairly happy with the results. It provides a simple, yet customizable enviorment, it’s fast and really simple. I’ve tried to create an editor that doesn’t get in your way and that actually tries to help you. It was built for my needs so, these are addressed completely. I might add more features in future versions if anyone requests them since, I want it to be good editor for everyone.

I have a small list of things I’ve been considering to add so, we’ll see how that ends up working. Hopefully I’ll get somewhere with it and, even if it’s not a good seller, I did enjoy building it, working with C# is entertaining and pretty cool so, it made the whole process a nice experience. If I don’t make a single sale for a couple of months I’ll open-source it since, it’d mean there’s no commercial potential for it.

PHP Routs

Mon, 11 Jul 2011 00:00:00 +0000

I am probably the most bored person I know. What I mean by this is, that I am almost never entertained. Most of the time, I'll be doing something for some 5 minutes before my short attention span takes me elsewhere. This has stopped me from finishing a lot of personal projects in the past, losing a lot of cool things I built halfway through.

Since, I almost never finish my projects, I've decided to start publishing them. Hopefully, just knowing that others are looking at my code my motivate me to continue developing something. I've finally started using my Github account to it's fullest. I honestly doubt it'll actually have a great effect on me but, any motivation I might get will prove useful. So, in conclusion, I'll be releasing most of the stuff I do whenever I am bored on my Github account and I'll be posting about it here, all in an attempt to keep me interested.

Introducing PHP Routs

Let's move on to my actual project. Routs, is a small PHP library built to handle URLs. Some background: I was playing with Sinatra the other day and, just by reading the documentation I fell in love with how it managed URL routes (I later discovered I wasn't the only since, there are a bunch of projects written in PHP following it's example). So, I decided I wanted to do the same in PHP but, in a very simple way which, would allow me to easily implement it into any project by including a single file.

It turned out to be a lot easier than I expected. After deciding on the syntax I would like to use with plugin I made huge advances in just a couple of lines. This is what I have so far:

Let's say I am building an simple one page site for some guy named Bob. Since it's so simple, I want to have as much as I can in a single page. Routs allow me to do something like this:

require 'routs.php';

Rout::get('home', 'homeFunction');

function homeFunction($params) {
  templateManager::Render('home.tpl', $params);
}

In this example I am using Routs to handle all get requests to example.com/home using the homeFunction() function. The part about the template manager is whatever method you're using, I'll be creating my own soonish. Now, I added a form to my html page and I want it to be handled on the same page. My routs handler changes into this:

require 'routs.php';

Rout::get('home', 'homeFunction');

function homeFunction($params) {
  templateManager::Render('home.tpl', $params);
}

Rout::post('home', function($params){
  mail(
    'seich@martianwabbit.awesome',
    'Some Subject',
    emailHandler::Render('mailTemplate.tpl', $params)
  );
});

Which handles the the submitted form and so on. Basically, the possibilities are endless! :D So yeah, that's it, have fun with it, fork it, tell me why it sucks, etc.

JustVector Social Icons Font

Mon, 04 Apr 2011 00:00:00 +0000

There's a new version available here.

I've been pretty bored lately, so I decided to redesign my site and found an awesome social icon set by Alex Peattie (http://www.alexpeattie.com) to go with the new design. The set is minimalistic and really cool! Unfortunately, I wasn't as happy to work with it as I would have liked to mainly because, I couldn't decide on what size I wanted the icons on the new site and thus, I found myself opening photoshop and resizing them far too often. I eventually, as you might have guessed by the post's title, decided that such an icon set would make an awesome font. So, I created a font file which included most icons from the set* . This would allow me to embed the font and get nice icons which, I can resize and recolor with css.

I am very happy with the results, I'll be updating it soon as I wanna get rid of some of the icons I included and replace them for others that I might think are more useful. The whole icon set is huge. So I'll keep all versions I make around just in case the icon you're looking for is in one of them. The main ones (Twitter, Facebook, Tumblr, etc.) will be on all versions of the set.

I've created a @font-face kit using fontsquirrel. So it's pretty much ready to be downloaded and used as it is right now. You can check out Alex's demos right here.

The font is distributed under the same license as the icons, the Free Art License which, allows you to pretty much do anything you want with it. Read the license if you want details.

* I didn't include all icons because, fonts have limited glyph spots available and I wanted to support as many keyboard layouts as possible by only including icons in common keys. For example, I didn't use any of the accent keys I have access to because I know most English keyboards wouldn't support them. (I have a Spanish keyboard)

Download Font Kit

Update: V1.2

I found the first font a little hard to navigate so, I created a second version in which icons are now arranged alphabetically (mostly) and empty spaces where no icons could go were filled in with additional icons. This font version has less glyphs than the first version because it only uses letters and numbers. If you need some of the additional icons you can use the first one.

Download Font Kit (v1.2)

Tips

*When you use these icons add a text-shadow. It'll make the images smoother. Something like: text-shadow: 0 0 1px #000; would do the trick.

Seichy\'s Missing Bit

Leica CL

Additional Documents

Restoring my Adana H.S. 2 6 x 4 Letterpress

Bookbinding Adhesives

Conclusion

We moved to Spain

Project Log: MidiFaders - ESP-32 MIDI-BLE Faders

Electrical Design

Milling a PCB

Firmware~

Case

The End ~

Making Softcases for Small Electronics and Cables

DIY Midi Cable for the OP-Z and ZM-1

Project Log: Some Sort of Guitar

January 22nd

January 24

Installing Gitea on Dokku

The Pandemic

Before The Pandemic

The first month

The following 4 months

The rest of 2020

A New Year

Installing FlatCAM on Ubuntu

Haproxy, Nginx and Dokku

KVM Backups, the easy way

PocketChip in 2020

Apt Sources

SSH Access

Checking the Battery Level

Using UART

A full desktop environment

Using your GPIO Pins

Configure the app launcher

Setting up Ruby on Rails System Tests on WSL with Docker

Notes on a Roland DXY-1150

The Power Supply

Communicating With The Plotter

DIP Switches

Conclusions

Glitch, The Cat

Restoring my HP7475A to it’s former glory

HPIB

Testing OCLIF apps with Jest

Hardware Debugging: Fixing My HP6236B

Beau 0.4

Writing Plugins for Beau

The Plugin

Conclusion

Hey Beau!

Making a Wooden iPhone Stand

Postgres: Checking Queries in Real Time

Expiring full-page caches in Rails 4 from the model

Angular.js One-way bindings

Read More!

JustVector Social Font 1.5

Using TableViewControllers with Swift

Setting up the table and cell prototype

The ViewController

Resources

Using MapKit with Swift

First, create a new project

Setup your desired location

Bonus: Adding an annotation

Additional Resources

Using mailgun to receive emails with Phabricator

Step 1

Step 2

Step 3

Sketch Presets for Desktop Screens

Scroll Spying for Fun and Profit

2014

You've Gotta Love Deferreds

Recommended links

A Semantic Grid with Zurb's Foundation

The Mixins

Side note:

Conclusion