[GSoC 2016] [Weekly Status] Capture Tracking Improvements

Hello all,

I will be writing brief weekly reports throughout the summer in a
similar style to other GSoC students in order to keep you updated in
how I'm getting on with the Capture Tracking Improvements.

This week I'm mostly been been focusing on research.

- I've looked at various papers as suggested on the mailing list.
- Looked through the current implementation for Capture Tracking.
- Looked into which optimisations use capture tracking. I'll be
continuing this next week as well.
- Looked at some of the existing test cases for Capture Tracking. I'll
be continuing this next week as well.
- Instrumented the current Capture Tracking implementation ready to run
code through in order to see what is marked as captured and what's not.

As always, any questions or comments are welcome.

Many thanks,
Scott

Hi,

Here's this weeks report.

- Began learning the LLVM assembly language.
- Compiled Firefox using an instrumented version of LLVM in order to identify what is currently marked as "MayBeCaptured".
- Analysed the result of this and looked for common trends.
- Began looking into analysed result in the algorithm to see if it appears to be being too conservative.

Many thanks,
Scott