My websites did not manage to dodge all of the bullets either. A few sites were down from the Media Temple thing. I also inadvertently broke a few sites here and there. After the Media Temple downtime, I wanted to try updating a few Ruby Gems to get Passenger Phusion working. Unfortunately my rails apps hadn’t been updated in forever, so updating things broke backwards compatibility. I’ve been working on getting Anime Nano back to full capacity on the newest version of Rails. I think the whole experience was a bit stressful, but probably ended up being good in the long run.

I was also surprised to see Mapskrieg going down late last week. Apparently I hadn’t verified the domain for Google Apps and Google decided to close my account. The one that was serving Mapskrieg via App Engine. This was really annoying since all of the “act now or we’ll delete your account” emails were classified as spam in gmail and I am pretty sure I had verified the account previously. If Google doesn’t want my money then screw them. I figure I can actually serve the site for cheaper than App Engine since a few other apps live on the same server. The App Engine thing was a fun experiment but there are still too many limitations in the datastore to base a real app on it. Specifically, datastore writes cost way too much and indexes take way too much overhead in disk space (which you eventually need to pay for).

So now I’m back to running most sites on Media Temple, though I may diversify in case Media Temple continues to have downtime issues. It was a stressful few days, but at least I got to practice my sysadmin skills a bit and update my server in the process.

It’s been about 4 years since I announced the launch of Mapskrieg, my Google Maps and craigslist mashup web app. Since then, I’ve gone to school (again), worked for Microsoft and quit, and made a few apps. Today, I’m happy to announce the launch of my newest app, Mapskrieg for iOS!

I’ve been working on this app for a few weeks, and I think it’s ready for public consumption. It’s basically Mapskrieg, but developed natively for the iPhone and iPad. In the past, I’ve taken a sort of iterative approach of releasing fairly minimal apps and improving on them. For example, I released Threadless as an iPhone only app and later added the iPad support. I wanted to release Mapskrieg on both platforms so the launch would have a little more bang. Plus I think the iPad app is the better of the two, and I really wanted that one to stand out for the release. I’ve been getting much more comfortable with mobile development, and I’m very happy with the rate at which I was able to conceive and release this app.

That’s not to say I didn’t struggle or learn anything new with it. While a lot of the concepts are borrowed from my Threadless iPad app, I had to do a lot of stuff I hadn’t done before. For example, since Mapskrieg is going to rely on iAds to make money, I decided that both the iPad and iPhone versions would support iAd. Apple, in their infinite wisdom, made the split view controller a very useful and now, standard, design paradigm. Unfortunately, they don’t provide any support for using it with iAd at all. What the fucking fuck, Apple!? So I had to basically recreate the Split View (well, the landscape mode at least) in order to support iAds. It’ll be well worth it if I can rake in some iAd dough, though!

This post also comes almost 1 year after I quit my job at Microsoft. That milestone probably deserves its own post, but I’ll just say that I have not yet regretted my decision in the very least so far. This is as fun as it gets, folks!

Edit: Oh, I forgot to link to a demo video that I recorded for a contest. Check it out in case you don’t have an iOS device:

I keep forgetting to mention this. A while back there was this cryptic ad on Facebook about mashups and it led to a Dice.com sponsored contest. I figured I would enter MapsKrieg and I sort of forgot about it. Checking the visitor stats, I noticed some people were coming from a contest poll page. So MapsKrieg was chosen as a finalist in this contest!

The prize is $4000, which is pretty neat. So please check out the finalists and vote (preferably for me)! Mine is #2, “Home & Apartments.” I wonder why they just didn’t label it “MapsKrieg?”

I did a bit more work on MapsKrieg (the official name of my craigslist and google maps mashup) in the past few days, and I think it’s ready for initial launch now.

If you haven’t read my previous posts on MapsKrieg, it’s a mashup that scrapes craigslist apartment listings and displays them using the Google Maps API. I’ve been on the hunt for an apartment in Ann Arbor, MI where I’m going to go to grad school. I had seen the coolness of HousingMaps, but I noticed that it only supported a very small number of cities. I figured I’d build a similar app that listed more cities, since craigslist has expanded quite a bit since that HousingMaps was created.

If you’re wondering about the name, here’s my thought process: I was originally calling the project “Craigsmaps” since that sort of made sense. But the domain name for that name is already taken. So I came up with a bunch of other stupid Web 2.0 names that didn’t really click (Abodr, apartment.icio.us (okay, just kidding)). So I switched the “craig” and the “maps” and it became “mapscraig” which reminded me of “blitzkrieg” which became “mapskrieg.” And there you have it.

Currently MapsKrieg supports 286 different craigslist locations. I haven’t added any areas outside the United States, though if I get requests, I’ll add them. And I’ll probably add new cities as their added to craigslist if people request them, too.

As far as technical stuff goes, I built the site on the CakePHP framework, which was a new experience for me since I had been doing stuff in Ruby on Rails for the past year. The app uses Google Maps’ API for displaying the data that’s pulled directly from craigslist listings. Of course, all the data is cached in a MySQL database, so the site is decently fast.

MapsKrieg took about two “man weeks” worth of time to complete. It was developed over about a month, though I didn’t work on it every day, and I didn’t work all day on the days I did work on it. I’d say that’s pretty good, considering I hadn’t really done much programming in PHP/Javascript before, and I had to learn the Google Maps API (which was surprisingly easy).

Anyway, go ahead and try out MapsKrieg and let me know what you think. The interface could probably use some tweaking, but I think the app “works” fairly well. But I’m open to any questions or comments.

I guess both of them like using the ‘$’ character as a shortcut to make the code easier to read. Which means they have crazy namespace conflicts. Bad developers! Anyway, here’s how I got it to work. Right after the import of the jquery library, I set the ‘$’ thing to be ‘$j’ instead. Then I refactored thickbox to use the ‘$j’ name instead. It worked. Here’s what the head looks like:


...
<script type="text/javascript" src="/js/jquery-latest.pack.js"></script>
<script>var $j = jQuery.noConflict();</script>
<script type="text/javascript" src="/js/thickbox.js"></script>
<script type="text/javascript" src="/js/prototype.js"></script>
...


This took a lot of time to do manually. I had to figure out if the area was an actual city, or a bunch of cities, then figure out the geo coordinates, and stuff it into the database. I thought it might be faster to do it by hand than write a script. Nope.

Today I wrote a script that would make the entry faster by automagically getting the coordinates from Google Maps’ http request service. It still took a while, but today I finished inserting almost all of the US cities (excluding the combination ones, since Google can’t do those).

Now all that’s left is for me to make the pop up windows a little prettier, and maybe add a list of the CraigsList listings to the side for easy browsing by price, # of bedrooms, etc. That shouldn’t take too long. Then I can launch the site!

I did a little research on the person who did the Google Maps Craigslist mashup like two years ago. Apparently he got a job at Google! While that’s pretty awesome, I can’t see this project doing the same for me, since it’s something that’s already been done (and two years ago, which is an eternity in web time). Rather, I’m just building this for my own use and perhaps for other people who wish there were more cities available in that HousingMaps mashup.

I came up with a name for the application. I was going to call it CraigsMaps, but someone already has that domain name. I’m thinking I should buy the domain name pretty soon, but what if I come up with a cooler name later!? The app should be ready in a couple of days anyway. Depending on how much extra stuff I want to cram in before I “launch” it.

The way that the custom controls work is both pretty sweet, and not that great. I don’t like having to add content to them by using javascript. It just doesn’t seem “correct” to do so, since I’m just overlaying normal html. So I’ll be using some custom controls, and some divs. The good thing about custom controls is that the map knows not to open infowindows under them when they’re opened up.

Besides that, I just did a lot of random work on getting the app to look a little better. Like showing the correct area based on which city’s listing you want to see. Overall, the project is coming along really smoothly, with very few roadblocks. I’m happy that it’s nearing completion. I really enjoy doing these one-man projects, since it lets me get experience in everything from database schema planning and optimizing, to php backend development, to designing the css for the content, and even maybe a little tweaking in the Apache server. I think once I add this project to my “real” server, I’ll need to get a ram upgrade.

I think the project should probably be ready to go live within a week. I dunno if I’ll actually try to get the word out, or just release it and see if people find it via Google or something. That’d be cool. Since there’s already a site that does a similar thing (though there’s a smaller number of supported cities), I don’t think what I’m doing is really that noteworthy. I think it’d be useful, though, so we’ll see if there’s any reaction at all to it.

I also did some stuff with the CakePHP helpers. CakePHP’s inferiority to Rails is really starting to show to me. I think it’s all a product of PHP having not been created for this kind of MVC design. CakePHP tries to follow the conventions of Rails, but it does so in a sort of copycat manner. As such, the result does not meet or exceed the original.

In the case of Helpers, Rails automagically links the named helper to the controller. In CakePHP, you must specify that a certain controller uses a certain helper. It’s kind of annoying. There also isn’t any support for the Rails convention of “partials,” which is why I’m using their helpers to begin with… It’s annoying.

Also, I find it annoying that CakePHP finds an objects associated objects by default when using “find()” or “findAll().” It’s useful when I want it, but when I don’t, it wastes a lot of memory! Now that I think of it, I’m not sure if Rails does this by default or not. The only reason I mention it is because I kept getting out of memory errors (or more accurately, memory limit errors) with my application. It’s a big waste when all I want is the name of the area and CakePHP fetches ALL of the associated listings for that area…

I must sound really whiny right now. I guess it’s just from some frustration in learning a new language/framework instead of using one that I’m used to. It’s good for me, I won’t argue against that. It’s just a bit disappointing to see all of these flaws when a framework is purported to be things like “stable” and “good.”

Oh, and also, I should say that I think my actual mashup is coming along pretty nicely. It should be ready in a few more “days,” where a “day” is one I am actively working on the mashup. I could probably “release” it pretty soon, but usually when I release something, it means I won’t work on it much anymore, moving to something else I find interesting. So I should probably actually have it done before I release it…

Besides that, I got the interface to what it was previously, except with AJAX so that the Craigslist images are only downloaded when you click on a map marker. I also found that it takes a long time to put a week’s worth of listings (from Ann Arbor, at least) onto the map at once. I’m guessing there were at least 400 on there. I made a hard limit to the number of listings that show up, and it loads a lot faster now.

I’m sort of back into the mood for working on this project again. I finally beat the Scrin campaign (it was really short) in Command and Conquer 3, so I’ll definitely be more focused now. I think.