Just a quick update so that this blog doesn't become too disconnected from the present. I had an amazingly quick job turn around after the end of Nexopia. Thanks to Boris Wertz providing some great introductions I connected with Weddingful where I started as "Technical Development Lead" on October 12th. I think it's an amazing opportunity, and I've been having a great time so far. It's great to get a bit more involved with macro company decisions. Hopefully I'll get the chance to write some more technical posts about the stuff we're doing in the not too distant future, but for now I'm headed to bed.

It has been a long time since I've written an update, but I've been meaning to start again since I moved to Vancouver. Having learned of Nexopia's planned wind down and my upcoming unemployment yesterday, I have the impetus to share all the exciting projects I've had the opportunity to be a part of over the last year. I'll take things in chronological order. My apologies if it's a bit long-winded, a year without updates leaves a lot of ground to be covered.

Project Chronos

Strange how a project name always seems cool when you are working on it and tacky after the fact. I finished up this project shortly after my last update. I took it to the point where it was a fully functional SMS reminder service, then Steph came home from Banff and other opportunities presented themselves. I think I accomplished what I wanted to with this project. It was my first attempt to do full stack development right from the server up to the graphics. I learned a lot and in the end had a working product.

Prototyper

Prototyper was/is a native OS X tool for making high fidelity prototypes of iPhone applications. Prototypes could then be viewed and interacted with through a web interface or directly on an iPhone/iPod touch. I first heard the idea for this from Cam Linke in November of 2009. He, Sean Healy, Sari Maani, and Sinan Maani were already on board and had identified a need for another developer to round out the team. I was sold pretty quickly on both the idea, and the chance to work with the four of them. My biggest concern with the project was scope, I felt we needed to execute quickly. I envisioned us being somewhere or being done within 6 months. We started getting together to do actual product and development work in about January 2010. We made friends with Trevor the manager of Original Joe's and he opened up the upstairs of the restaurant for us to work on Sundays. We had a great time working there, and in telling people about it ended up starting a regular little Sunday hackathon (#codejoes) where everyone would bring their personal projects and work for the afternoon. The project was exceptionally interesting from a technical standpoint for two reasons. It was my first time working with Cocoa, a surprising mental adjustment to make that has since influenced my development in other languages as well. We also had an interesting approach to our iPhone simulator. We really wanted to make it web-accessible so that it was easy to share prototypes and gather feedback. With that in mind we decided to do the simulator in javascript/html. We wanted a smooth and natural feel though, inertial scrolling, etc. To accomplish that we implemented two layers; one layer for rendering the prototype visually with a second  "glass" div overtop that captured all click/touch events and passed them onto the appropriate handler, handled scrolling friction and gesture support. Things were moving along pretty well, but scope was a problem. We were not getting enough development time into the project only working on it part time. We decided to apply to Y-Combinator, we felt we had a good team and a good idea. The decision day came and we got an email asking some follow-up questions about our work. They then extended their decision making process a day. Finally we got our answer and we were not in. Not a big shock given the huge number of applications they deal with, but a let-down all the same, particularly after they showed enough interest to ask some follow-up questions. We applied to a couple of other accelerator funds as well, but without success. There was talk of potentially raising some seed funding outside of an accelerator so that we could afford to work on things full time. I did some personal contemplation and considered the risks, market, my own financial situation, and decided I wasn't willing to go all in on this particular project. At the same time another opportunity was presenting itself to me in the form of Rival Apps, so in fairness to both projects I stepped back from Prototyper in about April of 2010.

Rival Apps

For the last four and a half years my full time job has been working for Nexopia.com. I came on during the original major growth stage, and have learned through all of the experiences that resulted in its eventual downturn. At the end of March 2010 the company decided to try to make a move in a new direction. We had been unable to turn around user decline on Nexopia and hoped to bootstrap a new opportunity while we still had the chance. Our fearless leader, Kevin Swan, in consultation with stakeholders, had been looking for a market opportunity and identified sports social gaming as an underserved market niche. I was asked to refocus myself away from side projects to lead development down this new road and I accepted the challenge. We started out with a small fantasy hockey draft app just to get a feel for integrating with Facebook and the tools available to us. We were abandoning 4 years on our custom Nexopia ruby framework in favour of Rails. We also moved to lighter processes and fewer in-house tools. I was able to apply some of my lessons learned from Chronos and Prototyper here. We quickly pulled Chris Thompson off of Nexopia to assist, and launched in time for the playoffs. The overall feeling was that we had accomplished our goals with the app - it wasn't a big hit but we had a couple thousand users and got a feel for what we were doing. 

Sadly, at this time Nexopia was forced to lay off about half of our remaining staff. All that was left on the product development front was refocused on Rival Apps for our next endeavour. Next on the list was a game we called Baseball Empire. The idea was to create a baseball simulation and management game where you combined your friends (who were trainable) and real players (who could only be contracted to play on your team for limited periods of time). We made one major technical decision early on that shaped a lot of the development, the adoption of Sproutcore. We made the choice to try Sproutcore because we had been dissatisfied with the Rails view layer. We wanted a lot more interactivity and javascript than was easily integrated there. We continued to use Rails for the server-side logic, but moved the entire view to Sproutcore, pulling data from the server via JSON objects. Sproutcore was an excellent tool for this (most companies choose flash for this role but we simply didn't have the expertise or the time to gain it even if we had been interested in flash), though did impose a learning curve on development. The Sproutcore framework is not incredibly mature and we did spend time debugging the framework as well as our own code. In the end I'm not sure if Sproutcore was the correct choice as it came with a high cost. Redeeming it was its binding layer, which made what is normally a complicated mess of spaghetti code to update the screen, simple and elegant.

 

Baseball Empire development has been going well, however where it will go from here is uncertain. We "launched" very early on and it has been rewarding to see the gradual growth in users as we've fleshed our product out, from something that could barely be considered a game, towards a fairly addictive app. I'll hold off on further commentary here until the future becomes clearer.

Green Planet

I attended the first Canadian Startup Weekend in Edmonton the last weekend in June 2010. It was an amazing time. I got to work with a great team of people, some of whom I'd work with before: Cam, Sean, Joel Jackson, and others for the first time: Ken Bautista, David Quail, Sam Jenkins, Estin Edwards, Mark Donovan. I swore going into the event that the one thing I wouldn't do was a Facebook app, but with a bit of convincing they brought me around and Green Planet was born. I was impressed with how much we were able to get done in a single weekend, and I had a blast doing it. If you are considering a Startup Weekend I highly recommend it. The next one I'm attending will be in Vancouver, October 8th-10th. Green Planet is a "social game with a conscience". You complete missions and then mark them as done in the game. This in turn causes your planet to become healthier. You can also view your "Universe" to see how healthy your friends' planets are. If we get the chance to work on it more in the future, your planet will also get unhealthier over time if you don't complete any missions. As well we would add check-ins for smaller day-to-day tasks you can complete. Along with the Facebook app we also managed to produce an iPhone app so that you can check in on the go, it has not been submitted to the app store however.

Vidclipper

This was primarily a gag project, we made a trip to Calgary's DemoCamp from Edmonton, and built a demo en route to present. Our team consisted of Sean, Joel, and myself doing development. Reg Cheramy was our idea man and acquirer of a free van rental from Rent-A-Wreck and Mark did the driving. We took our inspiration from Instapaper. The goal was to make a bookmarklet you could execute on any page with a video, and then have the video queued up for later viewing in an iPad app. The project started when we got into the van, we had four hours before our demo. Needless to say it was a bit of an insane scramble of hacked together code. In the end we reduced the feature down to only working for youtube videos as we could easily (we thought) grab the h264 versions of them. We had a last minute bug that was preventing videos from loading in the iPad. We weren't able to figure out what it was and scrambled as we sent our demoers up to show off the app thinking that it was broken. As they reached the part of their demo where they warned the crowd that we hadn't quite been able to get it all done and that the next thing wasn't going to work, a miracle occurred. Our demo spontaneously worked. For every time I've demoed something only to find it broken for no known reason the universe paid me back. It was a great trip and a fun little project. I don't know that I'll ever again agree to try and complete a product entirely on a car ride though.

Concluding Thoughts

The last year has been a great time. I have learned a ton, both on the technical side, and especially on the business side of startups. I'll never be the "business guy", nor do I want to be, but I have a much better appreciation of what happens there now. I'm sad that my time with the Nexopia team is coming to an end as the people there really helped shape the person I've become. I'm excited for the chance to start somewhere fresh though. I'm in a new city ripe with opportunity, looking forward to finding something I can help make great.

Been awhile since I updated, spent more time coding than writing about it I guess. Project Chronos is finally up and running in private alpha. Things are still quite preliminary, it allows you to create and edit reminders, which are then sent to you at the proper time via an SMS. Here are some screenshots to get a feel for it:

The reminder creation page lets you create new reminders, clicking on the clock or calendar brings up time or date selection widgets.

Once you create a reminder you are presented with a summary screen showing a list of your upcoming and recently sent reminders.

That's really about all there is to it so far. I have a lot of ideas about other things I'd like to add, but I want to get a few people trying it out before I make any firm decisions. If you'd like to help me out by testing it, and are willing to deal with potentially some bugs and changes, please drop me a line. You can get ahold of me (@LogicWolfe) via Twitter or any other means you prefer.

Been making steady progress on Project Chronos this week. There will be a new blog post here soon with some details on status.

The first is Hoptoad, it is a rails service for tracking application errors. It aggregates error, tracks what deploys they've happened with, and integrates nicely with Lighthouse and GitHub (in the pay versions). It also sends you a notification email the first (and only the first) time a new error occurs. The integration though isn't at all required to appreciate the app. It makes it incredibly convenient to track what types of things are going wrong, and how frequently they are occurring. We have some in house tools designed to do this kind of thing at Nexopia, but nothing as polished. The best part about it is from the time I decided to install it to the time I had completed integration was under 2 minutes.

The second cool new service I found is New Relic, recommended to me by @tvongaza after I mentioned my pleasure with Hoptoad. New Relic does performance monitoring for your app. Measures the time it takes to complete a page, the cpu load, database load, etc. It also has a very nice development mode that lets you drill down into the details of individual transactions on your development server. It has a free version that is available for unlimited hosts, but it doesn't include the full feature set of several quite expensive pay versions. The feature set of the free edition is still excellent, it can be a bit annoying to see the constant reminders of features you can't use though. Between these two excellent services and my recent adoption of GitHub, I am feeling pretty happy about my developer tools now.

I've been doing version control for project chronos with git. Since starting with git I've been looking for a good project management/source code repository site that was free. For awhile I was using project locker, they provided Trac integration but other than that had about the worst UI I've ever used. After a round of frustration trying to find the URI to clone my repository from, I decided it was time to look for an alternative. What I found was Indefero. It is also free up to 25MB of storage, has a relatively friendly if spartan UI, and is open source so if you'd like you can install it on your own servers. I was very impressed. I figured I'd roll with the hosted version until I hit my cap and then install it on my own server. Sadly it was lacking one key feature: multiple ssh keys per user. I use several computers, each has its own public ssh key, and I really don't want to go around synchronizing them all. I also don't want an account per computer. This was kind of a show stopper for me. I did some reading and it looks like a patch is available that overcomes this problem, and will presumably make its way into the hosted version at some point. I decided though, as is often the case, that it wasn't worth the time I was spending trying to get something for free. I've used GitHub in the past for open source projects. It is a wonderful product. Incredibly straight forward, it "just works". $7/month got me their micro plan which covers all of my current needs and I don't have to worry about what to do with my code anymore. If you're looking for something free and willing to spend some time on it I recommend Indefero. If you just want to get on to development and don't mind a small fee (or free for open source), go with GitHub.

 

Update: Source code available from http://github.com/LogicWolfe/FBJS-Time-Picker.

Another full day at ENTS developing Project Chronos, today I tackled the reminder creation UI. The facebooker gem has a handy helper class for generating forms in an FBML kind of way, this simplified the getting started a great deal. However once you move past the basic html input types it is no longer useful. My attempt to add a simple javascript calendar is where my troubles began. The most important design aspect of my reminder app is the simplicity and ease of creating reminders. That in mind the default option of 3 select boxes to choose a date is not a viable option. Javascript calendars are normally a dime a dozen on the web and I figured it'd be no problem to add one in. That is when FBJS punched me in the face. While I absolutely admire FBJS from the standpoint of a technical accomplishment on Facebook's part, it is a pain in the ass if you want to reuse any libraries (and who doesn't?). Facebooker has ported a lot of prototype to FBJS, which means a lot of normal rails helpers work. For the life of me though I couldn't find any drop in calendars that were compatible or easily ported. What I did find on a Facebook developer forum was a kind of hacky solution that I could start working on (oh so greatful to whoever s10sys is at this point). About this time Rob Davy invaded ENTS and I spent quite a bit of time over the remainder of the night chatting with him. What work I did get done was my first attempt at hacking into Rails at a frameworky level. It is entirely possible that such was not the appropriate spot for my solution, but it was effective. I added a custom helper in the Facebooker form builder class to generate calendar fields with the appropriate bits of fbjs attached, and by 10pm when I left I actually have a functioning calendar control. Next on my list is a time picker control. Since there are less graphical bits required and I've discovered how poorly most existing libraries play with fbjs I think I will just roll my own based on another solution I've seen and been intrigued by. It uses a simple 2 click interface to select any time. If anyone knows of any wonderful fbjs libraries that my googling failed to find, please drop me a comment. Assuming they don't exist, this seems like a huge gap waiting to be filled.

I spent all day at ENTS getting started on my new hobby project to build a facebook app. I work in Ruby at Nexopia so it seemed an obvious language choice. I have only dabbled in rails (we wrote our own web framework at nex), but I know enough to get started, so the big thing to figure out was Facebook development. I started with some extensive googling, with limited success. It seems there historically were two main Facebook gems for rails, RFacebook, and Facebooker. Facebooker as best I could determine seems to have taken over entirely, but its tutorial information seems largely broken and out of date (where it even exists). The API docs seem okay so far, but I really wanted some higher level getting started information. PeepCode to the rescue. They have a pdf and screencast available for $9 that provided an excellent starting point. With that information I quickly got my app setup on both Facebook, and a basic set of scaffolding in rails, then reached my next hurdle. Since I needed Facebook to talk to my app to test anything, I needed a public IP for my development code. There are several options to achieve this, and I tried most of them before finally settling. Developing directly off of my VPS was an option, particularly by using sshfs to mount the project files. The problem there is largely performance. Textmate in particular makes the assumption that you have faster file system access than is provided by sshfs. Opening up a local port could have worked as well, but I deal with enough different firewalls moving around that it just isn't really an option for me. That left ssh tunnelling, which I finally settled on. It's a bit awkward but it allows instant testing of changes, through Facebook, as you make them. I setup a local mysql server (since Facebook uses 64-bit UIDs Rails requires a special plugin for this, scaring me away from multiple database types), and forwarded port 3000 (the default rails test server port) to my remote server, configuring my Facebook app to look on the non-standard port. Tested a basic rails scaffolding through Facebook and rejoiced happily upon seeing all the default forms working in Facebook blue. I had hoped to get a bit further in actual development today, but given the awkwardness of getting started with Facebook and Rails I'm happy to be as far along as I am. It was fun hanging out at the ENTS space, met a couple people as they popped in throughout the day. Looking forward to going back tomorrow to start work on implementing a UI.

I went to Pub 1905 tonight after work to return GoW2 to Colin, and to have a couple beers. Ran into David there as well. Amidst yet another debate on religion, I decided to be antisocial and broke out my laptop. I have been wanting to make the switch from slicehost to rackspace cloud for awhile now and the dawn of a new project seems a reasonable time to make the switch. I setup a new server on rackspace and named it honeybrown in honor of my surroundings. A couple of beers and a burger later I found myself at ENTS configuring the new server. I've been doing some reading lately and for my purposes Phusion Passenger and Ruby Enterprise Edition seem an excellent way to reduce my memory footprint (always the sticking point for me with a 256MB VPS). I managed to get both configured with very few obstacles, many points to the Phusion guys for providing amazing installers. I updated my rails versions, setup a git repository through Project Locker, and generated a basic rails app. Project Chronos now officially exists, albeit as just the default rails stub. Tomorrow I hope to dig into actually getting some development done.