WG-Polonius meeting minutes: 2019/03/07

Relevant links:

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

The first meeting of the WG Polonius was dedicated to finding concrete next steps, both as part of working on the roadmap, and as onboarding @albins into the group. Since Albin will do a masters related to Polonius, we also worked out other areas they could tackle as time progressed.

Ultimately, a good starter issue was found in creating a new Polonius “hybrid” variant, bridging the location insensitive and complete analyses: a realistic way for rustc to use Polonius.

We then described a possible follow-up task, focusing on Polonius completeness: incorporating rustc’s move/overwrite analysis, in a way similar to how it was done in Lark.

We also spent time talking about next steps in documentation, whether in talks (Niko) or written docs (Rémy).

Possible talks:

• a walkthrough of a more complex theoretical example.
• a more practical walkthrough with a focus on the concrete parts of the Polonius computation: interfacing with rustc and transferring data, computing the analysis' datalog rules, how datafrog works, etc.

Following those, we also mentioned a longer term task: the more complicated analysis handling higher-ranked regions described in this blog post.

The final points were more about cross-cutting concerns applicable to all WGs: how would Github permissions work, and setting up the polonius and datafrog crates and repositories to follow the soon-to-be finalized “crate conformance policy”.

The major next steps we decided on were:

• Matthew will write mentoring instructions for the hybrid algorithm task.
• Niko and Albin will prepare the follow-up walkthrough of Polonius (the specific time, and contents).
• Niko will review the subset relation PR.