Mercy killing !

Almost a year since I last played around with www.biketowork.info
Killed it today.

After a year of dabbling around in web technologies, Ruby and Javascript I feel I have understood the basic concepts and web design a bit better.

SO, am in the process of rewriting it. Hopefully the new avatar should be out soon :)

Not slow...

Distance: J. P. Nagar III Phase to IISc - 17/18Km
Time taken: 55 minutes at 1730 hours on a week day in peak hour Bangalore traffic.
Transport: Hero thunder

It definitely takes longer on a motorcycle. Will post the route soon.

Circuits and Solder

I believe in keeping my cycle as light as possible. So apart from the base essentials there is nothing else. That means - no fenders, no bell, no lock, no lights and no reflectors.
Which can be kinda dangerous in the night (the no light and no bell bit). So, have been meaning to build a bicycle safetly light for quite some time now.

Sumukh n myself decided to build one. So got the circuit off the net, did some shopping for components at SP Road a lil bit of soldering, a few components soldered the wrong way and a bit of debugging.

And here are the results :-)


The circuit .... in the Dark

ps: Oh yeah, we also have a nascent group called DIY_Bangalore. A place where people who want to build stuff can meet, discuss and hopefully share ideas and motivate each other.

Tests !

checking in tests :-)

Finally got tests for all parts of the application now. I was not testing the controllers directly as info from the client was being sent to the server using JSON objects and I was not sure how I was going to test this. Just got this working B-)

Must say its much much better when you can implement a feature completely using tests before you even hit the UI. And as a 'side effect' you have the tests which will tell you when you break something by mistake.

Till Rajesh pointed out, I was building the application UI downwards. For example, saving of a route, I got the UI aspects (the markers, the Lightbox) working first and then put in the code for the actual saving on the server.

... a small change in thought ...

and this has changed our approach to building the application backend upwards.

Wanted to introduce the concept of a Road into the app. I wanted the ability to save info on the 'road' between two markers. So started from the server side. Got tests which checked for how Road objects were saved and brought up and their relationships (unit tests). Then started on what information the server would expect from users when they save a road and wrote tests around the saving of roads with that information (functional tests). Still no UI. But the feature is in place. Now just need to write up the UI and the whole thing plugs into place.

Beautiful

Its the best feeling

... when someone from your office comes up to you and says that they too want to cycle to work and make some enquiries about where they can get a good cycle.

Writing tests for Biketowork.info

Development on this site had come to a complete standstill. Sigh!

After a small chat with The Learning Journey man, and after all those fiascos trying to get stuff to work on the server decided to go ahead and TDD BikeToWork.info

So yesterday was spent in writing Unit tests for all the models that were being used for the site. And today was spent trying to get the existing UI stuff to fit on the sparking new models :D

That took some time, but looks ok now I guess.

Deleting a marker from the InfoWindow - FINALLY

Whew. That took some time.


deleting a particular marker

Clicking on the map places a marker and clicking on that marker opens up an Info Window (to capture information about the road that we added). The info window also has 2 buttons - one for saving the marker info and another for deleting the marker.

The deletion bit too ages. Finally got it working with some hacky code (yuck !).

The Bite !

The lack of tests bit me today :-(

All these days, I was not able to save my routes on the server and I had no idea why this was happening.

Today with a LOT of help from Pratik (the guys who is providing this awesome service called FreeOnRails, on which this app is hosted for now) I was able to FINALLY figure out what was happening, and nail the bug. Something that a simple test would have nailed right on, instead of us wading thru server logs and messing around with file permissions.

Whew. But on the bright side, got to have a chat with Pratik and get to know him. Hats off to this guy. If only more people would be like this. I mean, he spent like 2-3 hours just looking at the issues that I was having. Awesome, awesome. I was totally blown away by this. I didnt know him, and he didnt have to do this as he was not gaining any thing by helping me out. Hats of to you dude !

Finally !

a row in the database

Might not look like much. But to get that one row into the database, Rajesh and me had to manually muck around with the app for almost 2+ hours to figure out what was happening and why the application was not saving to the database (we were not able to get the loggin to work, so had to work on some error logs)

I guess this is where not having tests and not deploying fast enough leads to problems. I dont have any tests that check the sanity of the system (I felt that they would slow me down) and once I got the application working locally I didnt bother deploying the application onto the FreeOnRails server incrementally to see if it works. Just did a big bang deployment a few days back and was just not able to save the routes. (I cant imagine how huge products are actually coded up this way, without writing tests and not releasing fast enough)

Anyways. Into the early hours of the morning, we finally figured that it was some relationship issue.

Thanks Rajesh.