Meeting weekly 2012 09 25

Agenda

• 0.4
  • legacy-modes setting (nmatsakis)
• deletion of shape code (pcwalton)
• exports / pub / priv (graydon)
• relicensing (graydon)
• infrastructure (graydon)
• tell us all about strange loop / Emerging Languages conference, someone!

0.4

Issues targeted at 0.4: https://github.com/mozilla/rust/issues?milestone=7&page=1&state=open

• Niko(?): Demoding: Want to use + instead of ++ for scalars. Have patch, but currently leaks. Should this target 0.4?
• Graydon: Yes. An important selling point of 0.4 is "If you don't put modes in your code, you don't have to think about them".

Deletion of shape code

• pcwalton: new box annihilator works (surprising but great!)
  • last use of shapes is the log statement, should be easy to fix since %? doesn't use it
  • can kill shape code. we lose cycle collector but we weren't using it anyhow.
• general merriment

Exports / pub / priv

• graydon: implemented pub, pub use and so forth to mean "exports"
• graydon: what this means is that by default export keyword is meaningless
• graydon: you must add #[legacy_exports]
• nmatsakis: does it give a warning or anything if it sees the keyword export?
• graydon: no but not hard to do

Doc sprint

• graydon: brson is doing great work in this direction

Relicensing

• graydon: we seem to have reached a consensus with legal team
• graydon: systems library exemption on GPL2 means we don't need to worry about being used with GPL2 projects
• graydon: apache foundation and our own legal counsel agreed a terse 3-line ASL2 notice at the top of each file is sufficient.
• nmatsakis: including test files?
• graydon: unclear
• jesse: firefox tests (at least many of them) are public domain

Infrastructure

• graydon: yesterday we got sign-off to bring buildbot infrastructure into production
• graydon: not for 0.4 but during 0.5 sometime

Last-use

• https://github.com/mozilla/rust/issues/2633#issuecomment-8680306
• tjc: I made things compile but hit perf regressions
• brson: what kind of regression?
• tjc: I noticed that type checking got 3x slower, for example
• tjc: either we now having copies that used to be moves...
• pcwalton: ...that seems likely...
• nmatsakis: ...especially because of the fact that we permit impl. copies of ~[T] in rustc
• tjc: ok

Strange Loop

• dherman's slides: https://speakerdeck.com/u/dherman/p/rust
• there will be video: https://twitter.com/littlecalculist/status/250291385847123968