Wed Oct 7 Participants * Georgia * Ernest * Pradyun * Bernard * Sumana Agenda * Status & blockers * Pradyun: Yesterday, worked though surveys + project board. Wants help deciding order to do things in. Prioritization. * Georgia: No particular blockers. TODO: give Ernest a tweet. And: `ResolutionImpossible` error, backtracking message, clarifying what's needed. * Bernard: no real blockers. We have been getting more signups for UX Studies panel, and 20+ signups for "who uses pip?" * Ernest: there will be progress on minuting on python.org - https://github.com/python/pythondotorg/issues/1656 - want to be an alpha user? * Sumana says: yes! First iteration is specific to PSF board's needs, then, anything PSF is overseeing. * And: PSF liked our !YouTube video! https://www.youtube.com/watch?v=B4GQCBBsuNU * Sumana says: get Sept invoices in ASAP * Sumana: reschedule Tuesday meeting & approve invoices, wrangled issues with Pradyun to sort out 20.3 and 20.2.4, and working on meeting minutes * Pip backtracking & taking too long: info message https://editor.nuboso.ei8fdb.org/0S2BEZAUTZCJCv089n94Jw?both# * Pradyun: there are 3 broad things that need to happen * 1: what exact threshhold numbers we should use when presenting "we're taking too long" -- "8 backtrack choices" for instance -- threshold how many decisions we say "that was a bad choice" about * Georgia: how long (clock time) do n packages/backtracks take? Each one could take unknown amount of time. * Pradyun: we could use clock time OR # of backtracks as threshold. * Bernard: per notes from yesterday: tentatively decided "after 8 backtracks" for 1st message because 5-6 might be reasonable * Georgia: time expectations depend on the type of feedback you are seeing. Seeing SOMETHING changing/loading is an indication of progress. * 2: exact wording of message * 3. LATER: what nice things we can do once we have better output * Bernard: ideally we'd test this with users. we don't have time, so, going by discussion with maintainers, past data & experience. In messaging, give people a way to give feedback such as "this message is confusing" or "that was way too fast" * https://files.nuboso.ei8fdb.org/backtrack-messaging.mov -- 43 seconds. Thanks Pradyun for sharing a "prototype command line output" tool! Messages in https://editor.nuboso.ei8fdb.org/0S2BEZAUTZCJCv089n94Jw?both#Messages * Message 1 - displayed after X backtracks * INFO: pip needs to download different package versions to find a compatible set to install. This could take a long time. * DECISION: we'll stick to 8 for this message. * Message 2 - displayed after Y backtracks * INFO: This is a complicated process. It may not be successful - to abort press keys: ctrl + c * DECISION: we'll stick to 13 for this messages. * To improve how pip performs, tell us why this happened here: https://github.com/pypa/pip/issues/XXXX * Timeline: * Bernard: don't foresee any potential slowdowns/issues we'd have around this. * 20.2.4 and 20.3 planning - https://github.com/pypa/pip/milestone/38 milestone for 20.3 * we're going to put out 2 releases this month. * 20.3 -- second half of Oct. * This "shiny new resolver" release. * 20.2.4 -- now-ish. * documentation and performance improvements that we'd get out to users who're trying out the beta release. * needs commits/PRs to be cherry-picked onto 20.2.3 * Pradyun: this is accurate. We need to decide which exact performance & docs improvements to add in. Add to https://github.com/pypa/pip/issues/8511 . Exact commits/PRs we need to add in. * Outstanding questions * Georgia to follow up on the existing issues that haven't been updated yet. * Concerned about some overlap/tangling between current discussions around backtracking output and some other UX issues discussed earlier. * should we have a 20.3 beta/prerelease? * no specific reason -- Pradyun was uneasy about the changeover of flags (old to new default), is less so now https://github.com/pypa/pip/issues/8937 * UX next 3 months (notes from Georgia, as I imagine we may not get to this in the meeting today) * We are wrapping up survey & interview data collection in the next few weeks, and are analyzing those results to share in a useful way for the team and the community * From there we expect to have some recommendations overall, as well as clear resource needs for the project that will be basis for the trainings * We can share a larger overall update next week if that's helpful * Yes! That sounds good! * Prioritizing Pradyun's work * reviewing and merging PRs, mostly #8932 * cutting 20.2.4 * things to include: https://github.com/pypa/pip/issues/8511#issuecomment-701325392 * docs: #8942 (how to default to new resolver), #8807 (ux docs), and #8933 (index page about resolver and UX work) * additional docs suggestions: * https://github.com/pypa/pip/pull/8829 * https://github.com/pypa/pip/pull/8561 * https://github.com/pypa/pip/pull/8927 * https://github.com/pypa/pip/pull/7859 * https://github.com/pypa/pip/pull/8873 * https://github.com/pypa/pip/pull/8780 * https://github.com/pypa/pip/pull/8795 * performance: * https://github.com/pypa/pip/pull/8924 * https://github.com/pypa/pip/pull/8926 * https://github.com/pypa/pip/pull/8912 * https://github.com/pypa/pip/pull/8839 * implementing backtracking info message improvements * implementing work needed to flip the switch https://github.com/pypa/pip/issues/8937