Upstream specs #1240

andrykonchin · 2025-01-30T14:16:53Z

Scheduled specs changes upstreaming from CRuby/JRuby/TruffleRuby

Revert the part of commit 10917c5cc026f839a3dcd072b6e274eed211d0f7, "Update to 18032a7", that discarded the previous commit.

…e install` But we stub-out GEM_HOME variable for test-bundled-gems and others on ruby/ruby. It means the installation path mismatched with GEM_HOME variable always. We can't test this example collectly. ``` 1) Gem.bin_path finds executables of default gems, which are the only files shipped for default gems FAILED Expected File.exist? "/Users/hsbt/Documents/github.com/ruby/ruby/.bundle/gems/bundler-2.7.0.dev/exe/bundle" to be truthy but was false ```

…s class

…or#message

…dated condition on receiver name length

…osed connections

…bility

…y is removed

…checks Ruby 3.4 changes

…being"" This reverts commit f196750. These changes lead to failure a Windows job on ruby/spec's CI. The leaks detector reports that global variables are changed: ``` LeakError: Globals changed: {:verbose=>false, :debug=>false} to {:verbose=>nil, :debug=>false} ```

andrykonchin · 2025-01-30T15:15:27Z

@nobu Unfortunately I had to revert changes upstreamed from ruby/ruby to have ruby/spec CI passed (0ba6deb).

All the Windows CI jobs (e.g. this one) fail with the following error:

LeakError: Globals changed: {:verbose=>false, :debug=>false} to {:verbose=>nil, :debug=>false}

Wondering if there is another way to solve the issue on Windows. Actually I cannot reproduce any warnings on Windows (I tried on GitHub Actions though).

I've tried to discuss the issue after the last specs syncing and asked in ruby/ruby@99d5e6a#r150964722.

These specs fails with error: ``` Errno::EINVAL: Invalid argument - recvmsg(2) ``` on calling the following methods: ```ruby client.recvmsg(10) # or client.recvmsg_nonblock(10) ```

…emoved The following error is raised: ``` Dir#chdir does not raise an Errno::ENOENT if the original directory no longer exists ERROR Errno::ENOENT: No such file or directory @ dir_chdir0 - D:/a/spec/spec/rubyspec_temp/5588_0/27-testdir1 D:/a/spec/spec/core/dir/chdir_spec.rb:192:in `chdir' D:/a/spec/spec/core/dir/chdir_spec.rb:192:in `block (4 levels) in <top (required)>' D:/a/spec/spec/core/dir/chdir_spec.rb:191:in `chdir' D:/a/spec/spec/core/dir/chdir_spec.rb:191:in `block (3 levels) in <top (required)>' D:/a/spec/spec/core/dir/chdir_spec.rb:129:in `block in <top (required)>' D:/a/spec/spec/core/dir/chdir_spec.rb:128:in `<top (required)>' ```

eregon · 2025-01-31T12:44:48Z

It's probably related to ruby/win32ole#33
Maybe there can be a more local solution where suppress_warning do ... end is used just around the deprecated constants and/or adapt the specs to the non-deprecated constants if they are available since long enough?

eregon · 2025-01-31T12:46:24Z

@nobu @andrykonchin (and I) don't have a Windows dev machine so it's difficult for him to fix it. Could you try to fix it in a way which doesn't permanently modify $VERBOSE and only suppress warnings in small scopes?

nobu · 2025-02-01T09:55:57Z

I moved the top level constants for Win32OLE to under Win32OLE module, and deprecated the old names then.
It is OK of course if there is no warnings.

eregon · 2025-02-02T12:29:18Z

I don't see any warnings on https://github.com/ruby/ruby/actions/runs/13096898780/job/36540234336#step:22:989
Do you see warnings locally when running win32ole specs with the current master of ruby/ruby?
gems/bundled_gems does have win32ole 1.9.1 https://github.com/ruby/win32ole which should have the deprecations.

eregon · 2025-02-02T12:36:18Z

In any case, I think there are two good solutions if you see the warnings:

In fixtures/classes.rb declare the new constants if they are not there already, like WIN32OLE::RuntimeError ||= WIN32OLERuntimeError and use the new constants in all specs. Or just the latter if the new constants exist since long enough to pass specs on Ruby 3.0.
Reconsider Warning[:deprecated] = true in MSpec, but I think the comment on it makes sense and we should keep it as it is.

eregon · 2025-02-05T13:54:48Z

library/socket/basicsocket/recv_nonblock_spec.rb

@@ -113,6 +113,7 @@
      end

      ruby_version_is ""..."3.3" do
+        quarantine! do # May fail with "IO::EAGAINWaitReadable: Resource temporarily unavailable - recvfrom(2) would block" error


I would say retry on IO::WaitReadable for client.recv_nonblock(10), it just hasn't seen the closed socket yet.

Done in ruby/ruby#12710

Have no clue why these new specs fail on Windows - they raise Errno::EINVAL: Invalid argument - recvmsg(2)

ccc72c7

andrykonchin and others added 30 commits January 30, 2025 14:56

Quarantine failing new specs that require investigation

9a8009f

Reapply "Suppress WIN32OLE deprecation warnings for the time being"

f196750

Revert the part of commit 10917c5cc026f839a3dcd072b6e274eed211d0f7, "Update to 18032a7", that discarded the previous commit.

Skip examples related with OpenStruct in ruby/spec

5a90de7

[Feature #6012] Extend source_location for end position and columns

25162ad

Restructured irb related example at spec/ruby

8324eab

Prefer uname -n over hostname. (#12647)

ad4f652

Add fallback for hostname if uname isn't available. (#12655)

681b728

Handle environment where GEM_HOME is not available

04e354e

Add rb_data_define() C function

0d88403

Refactor rb_io_open_descriptor()

7041294

Add Range#overlap?

8d4c32d

Add missing specs for NoMethodError#message

b10aaf1

Fix NoMethodError#message when receiver is an instance of an anonymou…

deae4e6

…s class

Always use Module#name for non-anonymous module to create NoMethodErr…

ba7f17c

…or#message

Refactor Truffle::ExceptionOperations.receiver_string and removed out…

eb95409

…dated condition on receiver name length

Socket #recv* methods return nil instead of an empty String on cl…

1cffa0d

…osed connections

Fix Marshal.dump when a Float value is dumped repeatedly

c8c3fd4

Dump Range instance variables in proper order

1158caf

Add specs for Marshal.dump and objects table

91a8e2c

Add specs for Regexp used in REXML that was affected by ReDoS vulnera…

31c5d66

…bility

Fix Proc#clone and call #initialize_clone instead of #initialize_copy

3f92a62

Refactor specs for Dir.for_fd and use a mock instead of Object.new

2c1323c

Refactor specs for Dir#chdir

981da7b

Fix a spec for Dir#chdir that checks a case when an original director…

a37d52d

…y is removed

Fix validation of seconds in utc offset in String format

9e1b2a2

Use ruby_version_is instead of ruby_bug for a spec for String#% that …

c61df54

…checks Ruby 3.4 changes

Fix Integer#/ with bignum argument

0851dbd

Add Module#set_temporary_name method

b4f1a91

Add more Module#set_temporary_name specs and fix edge case

3270443

andrykonchin added 4 commits January 30, 2025 15:44

Add missing Ruby version guards for rb_data_define()'s specs

03ffa51

Quarantine some new specs for Socket#recvfrom_nonblock()

fbbb7d0

Fix Rubocop's warnings

d4fccb6

andrykonchin added 2 commits January 30, 2025 17:19

Skip some new specs for reading from a closed socket on Windows

ccc72c7

These specs fails with error: ``` Errno::EINVAL: Invalid argument - recvmsg(2) ``` on calling the following methods: ```ruby client.recvmsg(10) # or client.recvmsg_nonblock(10) ```

andrykonchin force-pushed the ak/upstream-specs-30-01-2025 branch from 8126f9f to cbbb3c8 Compare January 30, 2025 15:21

andrykonchin merged commit affef93 into master Jan 30, 2025
34 checks passed

andrykonchin deleted the ak/upstream-specs-30-01-2025 branch January 30, 2025 15:31

eregon reviewed Feb 5, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Upstream specs #1240

Upstream specs #1240

andrykonchin commented Jan 30, 2025 •

edited

Loading

andrykonchin commented Jan 30, 2025 •

edited

Loading

eregon commented Jan 31, 2025

eregon commented Jan 31, 2025 •

edited

Loading

nobu commented Feb 1, 2025

eregon commented Feb 2, 2025

eregon commented Feb 2, 2025

eregon Feb 5, 2025

andrykonchin Feb 7, 2025

andrykonchin Feb 7, 2025

Upstream specs #1240

Upstream specs #1240

Conversation

andrykonchin commented Jan 30, 2025 • edited Loading

andrykonchin commented Jan 30, 2025 • edited Loading

eregon commented Jan 31, 2025

eregon commented Jan 31, 2025 • edited Loading

nobu commented Feb 1, 2025

eregon commented Feb 2, 2025

eregon commented Feb 2, 2025

eregon Feb 5, 2025

Choose a reason for hiding this comment

andrykonchin Feb 7, 2025

Choose a reason for hiding this comment

andrykonchin Feb 7, 2025

Choose a reason for hiding this comment

andrykonchin commented Jan 30, 2025 •

edited

Loading

andrykonchin commented Jan 30, 2025 •

edited

Loading

eregon commented Jan 31, 2025 •

edited

Loading