WG-Polonius meeting minutes: 2019/04/23

Relevant links:

• The meeting happened in this Zulip thread
• A Paper document for the WIP roadmap

In this meeting, we started with a short review of the previous week’s work:

• Albin continued work on implementing liveness in Polonius, and finalized #105, which is now ready to be merged into a new polonius-engine release.
• Niko wrote up notes describing the second part of the liveness effort: modifying rustc to emit var_used and var_defined facts for the previous PR. Albin has already started working on these rustc steps.
• Rémy landed the tiny #106.
• As part of the Hybrid variant, Rémy started prototyping a low-hanging performance task: using the LocationInsensitive potential errors results to filter some of the input relations of the DatafrogOpt pass.

We then welcomed a new contributor, @lokalmatador, and spent some time discussing a good first task: using rustc-perf’s benchmarks, and Albin’s microbenchmarks, continuing the work towards an extended performance test suite. The goal would be to find crates, and functions, which would be representative along different dimensions, and incorporate their facts files in the Polonius repo to be used for tests and benchmarks. Niko described the basic steps in the Paper document. Link to full discussion.

A useful task was mentioned for the near future: going through the Polonius failures in rustc’s test suite (using the --compare-mode polonius test mode), categorizing the failures, and possibly making the tests pass, by --blessing the expected polonius test output files.

For the following week:

• Albin will continue on the liveness fact generation in rustc.
• Rémy will continue work on the Hybrid variant optimization.
• @lokalmatador will familiarize with the steps Niko wrote, with available help from the rest of the WG.