Skip to content
This repository has been archived by the owner on Oct 9, 2018. It is now read-only.

Latest commit

 

History

History
80 lines (54 loc) · 3.29 KB

meeting-2016-01-05.md

File metadata and controls

80 lines (54 loc) · 3.29 KB

Core team 2015-01-05

Agenda

  • RFC 1214 (nmatsakis)
  • Assorted trademark issues (aturon)
  • FAQ review planning (brson, aturon)
  • Backcompat issues (pcwalton)
  • PR reviewing backlog (aturon)
  • One-off contributors (aturon)

RFC 1214

Link: https://github.com/rust-lang/rfcs/blob/master/text/1214-projections-lifetimes-and-wf.md

https://gist.github.com/brson/5165e7f4dd74e5403045 <-- raw regr. report https://internals.rust-lang.org/t/regression-report-stable-2015-12-10-vs-nightly-2015-12-21/3026 <-- internals thread https://tools.taskcluster.net/task-inspector/#UkIbvu99QNSzMCQVwfiexw/0 <-- diesel

  • brson: Seem to be a lot of regressions around RFC 1214
  • brson: Looks like the biggest breakage we've seen, by far
  • aturon: But we've had warning cycles, right?
  • nmatsakis: multiple
  • aturon: Surprising that people haven't updated?
  • nmatsakis: Not really. But not clear how many of these crates are active.
  • aturon: It's going to an error only on current nightly?
  • nmatsakis: Yes
  • aturon: Maybe we should revert until we understand better the situation.
  • nmatsakis: Maybe, but I'd like to have a clear plan about when it's OK to roll out.
  • aturon: How much outreach did we do for this?
  • nmatsakis: Not a lot.
  • nmstaskis: Reluctant to roll back because there's a lot of complexity in the compiler with the warnings -- it's actually hard to roll back.
  • nmatsakis: Alternatively, can introduce a lint with default to error rather than warn -- that addresses issues with upstream crates, due to the lint cap.
  • aturon: I like that a lot. Separates breakage into purely local pain, rather than amplifying it through dependencies (and resulting in dependency hell)
  • nmatsakis: It doesn't look like most of these regressions are tied to RFC 1218

General plan: in the future, warnings -> errrors by default (but overridable) -> hard errors

Action items:

  • assess how many regressions are tied to RFC 1218 (nmatsakis, ASAP)
  • get the message out about the upcoming breakage (beyond RFC 1218) -- brson to cross-post, after regressions are triaged

FAQ

rust-lang/prev.rust-lang.org#202

  • brson: A bunch of web page work is about to land. The FAQ is one of the most important. Need to get a lot of eyes on this.

Backcompat issues

https://www.reddit.com/r/rust/comments/3ylgvf/why_are_projects_from_just_a_few_months_ago/

  • pcwalton: There was a reddit thread discussing backcompat issues (many people chiming in). Three main categories:

  • #1: libc was reexporting a lot of winapi; it was fixed in some release

  • #2: std::os::errno became non-public

    (but this was never stable?) http://doc.rust-lang.org/1.0.0-alpha/std/os/fn.errno.html removed in 1.0.0-beta

  • #3: intrinsics became private

    (but they were never stable...)

  • #4: std::libc was public but went private

    (?? was removed in Feb 2014)

  • #5: some crates where having multiple versions is deeply problematic, e.g. log

http://doc.rust-lang.org/1.4.0/std/os/index.html http://doc.rust-lang.org/1.5.0/std/os/index.html

Conclusion: perhaps we need to hide pre-1.0 crates by default