Skip to content

Latest commit

 

History

History
61 lines (44 loc) · 3.1 KB

workshop-guidelines.md

File metadata and controls

61 lines (44 loc) · 3.1 KB
Raw

Workshop sessions

The LEWG is experimenting with using "workshop" sessions to discuss each paper in a small group before bringing it before the full working group. Groups of 3-6 people can have a conversation about a paper without needing formal methods for picking the next person to talk, and if they rathole, they waste fewer people's time.

Guidelines for workshop groups when reviewing a paper

Claim a paper to discuss by moving it into your group on https://etherpad.net/p/c++-lewg-[meetinglocation].

Select someone who will present the small group's discussion to the full group. This can be the same person for each paper, or a different person.

Also select someone to take notes into the issue pages on the wiki, so there's a permanent record of your discussion.

Your goal is to get as many papers as possible into a shape that the full LEWG should spend time on them. You have lots of options when reviewing a paper:

  • Point out technical problems and iterate with the author to fix them.
  • List contentious points in the paper so LEWG can focus on them. Try to identify points that anyone on the committee might find contentious. Don't try to resolve this sort of issue in your workshop group. However, if your group happens to have consensus on an issue you think others will find contentious, point that out.
  • Do a line-by-line review of the wording in the paper, to make sure that the paper's rationale clearly explains what the wording is trying to do, and that the wording accurately reflects the design.
  • Recommend that LEWG reject a paper.
  • Recommend that LEWG forward a paper to LWG. See https://github.com/cplusplus/LEWG/blob/master/paper-checklist.md for goals before we forward a paper.

Look at the previous discussions of a paper, to try to be consistent with previous LEWG straw polls on the issue. The issue should link to most previous discussions, but there may be some on the mailing list too. The wiki may also include comments from people who reviewed the paper before your session, but who aren't in the session.

On having interested people attend workshop groups

Try to get experts on a paper to attend your group while you're discussing it, by announcing your discussions on the meeting IRC channel (irc://chat.freenode.net/##C++-meeting). Try to include:

  • The author of a paper
  • People cc'ed on the paper's issue on https://issues.isocpp.org/
  • People who've done in-depth reviews on the mailing list before the meeting

However, not everyone who's interested in a paper should expect to be present, especially for important papers. (Having everyone present would make the group large, which removes the benefit of doing an initial review in a small group.) Instead, trust that someone in the small group will represent your opinion, or that you'll be able to express it when the paper comes in front of the full group.

Before the meeting

  1. The chair should run a prioritization poll on http://civs.cs.cornell.edu/ to figure out which issues are most important.
  2. Add the issues in order of importance to https://etherpad.net/p/c++-lewg-[meetinglocation], so workshop groups can claim them.