Misc #18977: DevMeeting-2022-09-22

• Status: Closed
• Assignee: -

[mame (Yusuke Endoh)]

The next dev meeting

Date: 2022/09/22 13:00-17:00 (JST)
Log: https://github.com/ruby/dev-meeting-log/blob/master/DevMeeting-2022-09-22.md

• Dev meeting IS NOT a decision-making place. All decisions should be done at the bug tracker.
• Dev meeting is a place we can ask Matz, nobu, nurse and other developers directly.
• Matz is a very busy person. Take this opportunity to ask him. If you can not attend, other attendees can ask instead of you (if attendees can understand your issue).
• We will write a record of the discussion in the file or to each ticket in English.
• All activities are best-effort (keep in mind that most of us are volunteer developers).
• The date, time and place of the meeting are scheduled according to when/where we can reserve Matz's time.
• DO NOT discuss then on this ticket, please.

Call for agenda items

If you have a ticket that you want matz and committers to discuss, please post it into this ticket in the following format:

* [Ticket ref] Ticket title (your name)
  * Comment (A summary of the ticket, why you put this ticket here, what point should be discussed, etc.)

Example:

* [Feature #14609] `Kernel#p` without args shows the receiver (ko1)
  * I feel this feature is very useful and some people say :+1: so let discuss this feature.

• It is recommended to add a comment by 2022/09/19. We hold a preparatory meeting to create an agenda a few days before the dev-meeting.
• The format is strict. We'll use this script to automatically create an markdown-style agenda. We may ignore a comment that does not follow the format.
• Your comment is mandatory. We cannot read all discussion of the ticket in a limited time. We appreciate it if you could write a short summary and update from a previous discussion.

[mame (Yusuke Endoh)]

• Related to Misc #14770: [META] DevelopersMeeting added

[mame (Yusuke Endoh)]

• [Feature #18784] FileUtils.rm_f and FileUtils.rm_rf should not mask exceptions (mame)
  • I changed FileUtils.rm_f/rm_rf so that it raises a SystemCallError but Errno::ENOENT.
  • But it seemed to affect mkmf on Windows. mkmf uses rm_f to delete conftest.exe, but sometime it fails. I don't know why it happens.
  • I concern that this change will make users just ignore exceptions of FileUtils.rm_rf by wrapping begin .. rescue SystemCallError; end.
  • This is not what I wanted. Should we revert the change of FileUtils?

[byroot (Jean Boussier)]

• [Feature #18982] Add an exception: false argument for Queue#push, Queue#pop, SizedQueue#push and SizedQueue#pop (byroot)
  • The existing exceptions (ThreadError) are a bit awkward to deal with.
  • Queue is often used deep into the stack where exceptions in common cases are fairly costly.
• [Feature #18951] Object#with to set and restore attributes around a block (byroot)
  • Works with any public accessor, e.g. GC.with(stress: true) { do_something }
  • This is an extremely common pattern, especially in test suites, but is often implemented with subtle bugs.

[Eregon (Benoit Daloze)]

• [Feature #11689] Add methods allow us to get visibility from Method and UnboundMethod object. (eregon)
  • I think we should re-add (undo revert) {Method,UnboundMethod}#{public?,private?,protected?} on master, see https://bugs.ruby-lang.org/issues/11689#note-27
  • Both #18729 and #18751 are fixed, and now Method == method entry, and visibility is an attribute of method entry.
  • Ruby 3.1 already has those method and the fix for #18729 and #18751 was backported to 3.1. So if we keep them removed for 3.2 we cause incompatibility and there is no practical issue with those methods anymore.

[st0012 (Stan Lo)]

• [Feature #18996] Proposal: Introduce new APIs to reline for changing dialog UI colours (st0012)
  • irb's autocompletion background can make texts hard to read for some users.
  • Because the colours are hardcoded in reline, users aren't able to change them. So many users need to disable the feature altogether.
  • The APIs proposed in the ticket actually have been implemented by @pocari and me.
  • Since the maintainer @aycabta (aycabta .) is not available now, I hope we can decide if the APIs still need any improvement or they are ready for release.
• [Feature #19010] Follow up of #18996: Support changing irb's autocompletion background (st0012)
  • If the above or similar reline APIs are introduced, I want to support dark/bright themes in irb (screenshots are included in the ticket)
  • The default will be dark theme. I don't want to add more themes for now. So it can be toggled to bright theme with a boolean config.
  • To customize more, irb can take and pass per-color configurations to reline too.

[Eregon (Benoit Daloze)]

• [Feature #18798] UnboundMethod#== with inherited classes (eregon)
  • OK to change UnboundMethod#== to check if same method definition (and ignore from which class instance_method was used on)? If yes, what about UnboundMethod#eql??
  • If not, OK to add {Method,UnboundMethod}#same_definition?(other)?

[jeremyevans0 (Jeremy Evans)]

• [Bug #18960] Module#using raises RuntimeError when called at toplevel from wrapped script (jeremyevans0)
  • Do we want to support this? I think since #using is permitted at top level and inside a module, we should.
• [Bug #18919] Ractor: can't share #Method objects (jeremyevans0)
  • Is this a bug? Ractor cannot share most objects, but some procs are sharable.
  • If not a bug, should it be switched to feature request?
• [Bug #18878] parse.y: Foo::Bar {} is inconsistently rejected (jeremyevans0)
  • This is currently invalid syntax, should we try to support it?
  • If so, do we consider this a bug or a feature request?
  • I tried to support it a couple weeks ago, but all my attempts ended in reduce/reduce conflicts.

[shioyama (Chris Salzberg)]

• [Bug #18960] Module#using raises RuntimeError when called at toplevel from wrapped script (shioyama)
  • @jeremyevans0 (Jeremy Evans) mentioned this, but I think this is a bug based on common reading of the wrap argument to load.
  • See also context for this in my next bullet point.
• [Feature #10320] require into module (shioyama)
  • Proof of concept implementing imports like CommonJS etc using some patches to extend top_wrapper: https://bugs.ruby-lang.org/issues/10320#note-13
  • Key point is that I want to support existing code, including gems, with no changes.
  • POC gem: https://github.com/shioyama/im, requires patched Ruby
  • This is just for discussion of feasibility/approach.

[peterzhu2118 (Peter Zhu)]

• [Bug #18978] Unexpected behaviour in Time.utc and Time.local when 8 arguments are passed in (peterzhu2118)
  • Passing 8 arguments causes both the 7th (microseconds) and 8th arguments to be ignored.
  • The behaviour is undocumented, so it is inconsistent between various Ruby implementations.
  • Proposed fix raises an ArgumentError when 8 arguments are passed in (this is the behaviour when 9 arguments are passed in).
  • Proposed PR: https://github.com/ruby/ruby/pull/6281

[byroot (Jean Boussier)]

• [Feature #18885] End of boot advisory API (byroot)
  • The general concept was accepted, but it need a proper name.
  • Suggestions:
  • Kernel.booted / Kernel.application_booted / Kernel.code_loaded / Kernel.startup_done
  • ObjectSpace.loaded / ObjectSpace.optimize
  • The suggestions try to describe an "event" or point in time, more than which optimizations are being performed.
  • This is because the actual optimizations might change over time or be different on other implementations.
  • I'd like to implement the constant cache pre-warming soon, so having an accepted name for it would be very helpful.

[tenderlovemaking (Aaron Patterson)]

• [Feature #18776] Object shapes as new caching system
  • YARV interpreter benchmarks are similar speed or faster
  • Memory increase is low
  • Opens door for more optimizations

[ioquatix (Samuel Williams)]

• [Feature #19008] Introduce coverage support for eval.
  • Working implementation.
  • Can introduce some challenges for existing code - same challenges as debug gem - accurate line information for eval must be given.
  • Useful for computing template coverage.
• [Feature #18411] Introduce Fiber.blocking for disabling the scheduler (efficiently).
  • Useful for writing pure Ruby schedulers.
• [Feature #18455] IO#close has poor performance and difficult to understand semantics.
  • We need to fix implementation of IO#close which is O(number of blocking operations).
  • We need to hide implementation of struct rb_io_t because the internal implementation details are leaking into public interface (e.g. ccan_list).
• [Feature #18630] Introduce general IO#timeout and IO#timeout= for blocking operations.
  • We can now support generally blocking and internally non-blocking operations (i.e. fiber scheduler).
  • Safer network I/O.
• [Bug #18886] Struct aref and aset don't trigger any tracepoints.
  • Do we care enough to fix this issue or is the performance cost too high?

[yui-knk (Kaneko Yuichiro)]

• [Feature #19013] Error Tolerant Parser
  • I want to get feedback for the direction and other use cases.

[mame (Yusuke Endoh)]

We could not discuss all agenda items in time. We will have an extra meeting at 6th Oct.

[pkubaj (Piotr Kubaj)]

• [Feature #19022] Use __builtin_ppc_get_timebase on POWER with clang (pkubaj)

  • A simple change that allows run performance improvement

• [Feature #19023] Enable riscv64 coroutines on riscv64-freebsd, arm32 on arm*-freebsd and ppc on powerpc-freebsd (pkubaj)

  • Passes make test

[hsbt (Hiroshi SHIBATA)]

• [Bug #19016] syntax_suggest is not working with Ruby 3.2.0-preview2
  • What's blocker now?

[mame (Yusuke Endoh)]

• Status changed from Open to Closed

[mame (Yusuke Endoh)]

• Description updated (diff)