[Thali-talk] Yaron's Weekly Update - 9/2/2014

[Yaron Goland]

Last Week:

  *   TL;DR - After much bug stomping the first end to end IM demo worked on Thali!!!!! This is the first time we could use QR Codes to exchange identities, use Tor to communicate and have an app actually work via HTML end to end!
  *   Monday - Bug hell. I tried to find good solutions for handling getting logs during the gradle build so I could repro some issues and there really isn't anything great. I'm getting odd start up errors in Tor. I upgraded everything to the latest Tor binaries and that didn't really fix it. I ran into a weird android bug where it was complaining I hadn't registered a receiver. I finally had to just try/catch the error. I added in GeoIP6 support in Tor for no apparently reason. I fixed a bug in our build file for Android for Tor where we weren't copying files correctly. But the big problem of the day is that Tor just bloody well wouldn't start up! Making the wait times even more forgiving seemed to handled it. I also ran into an error where both the client and server have confirmed SOCKs connection no actual bytes will flow! So I had to harden the test to time out and retry when this happens.
  *   Tuesday - Spent a good chunk of the day running down a stupid test bug. I wasn't properly tearing down a new listener I added to the tests. I didn't see this because it doesn't matter on Java (where due to a Couchbase lite bug we can't tear down) and on Android it will only show up when running multiple tests. I wrote the HTML sample code for the IM app but am running into odd issues with resolving onion addresses.
  *   Wednesday - Ran into a scary bug where I'm getting a null pointer exception of line 232 of Document.java in Couchbase. This shouldn't be possible but it's happening. I repro'd at least twice and it crashes the Android TDH. At worse I can try/catch it but the real problem is that it means something is seriously wrong in the Couchbase internals. Unfortunately I have not been able to repro it too often. I then ran into a race condition in Android where the HTML can start before the relay or TDH are available so I wrote a javascript function to confirm when the system is running. I also ran into a really scary bug that I haven't fully repro'd where sometimes the relay just plain refuses to resolve onion addresses and the errors come back fast enough that it's clear it's the relay and not the TDH. Unfortunately the bug kind of went away which is itself scary since it means (like the null exception pointer) that it could come back at any time. But the good news is that eventually I did get everything working!!!

This Week:

  *   I need to bang on the demo to see if any of the bugs we ran into will repro and just to get it to the point where it's reliable.
  *   Publish all of our jars/aars to Maven and build our binaries on all of our platforms.
  *   Actually download a clean windows VM and go through all the steps a dev would have to go through in order to develop on our platform.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist10.pair.net/pipermail/thali-talk/attachments/20140902/6e4776be/attachment.html>