Hello again, all:
Now with again I tried to rebuild all rubygem-foo pacakages on Fedora with ruby 3.3.0dev:
The packages failing to build with: https://github.com/ruby/ruby/commit/9cd086ba4b559153864ab924723a665a4ddfb5d8 are listed below. (Commented inline between my previous post.)
8 packages are still failing to build.
Mamoru TASAKA wrote on 2023/10/29 17:09:
Hello, all:
With wonderful work by Vít and friends, again I tried to rebuild all rubygem-foo packages on Fedora against ruby 3.3.0dev.
For the below failure reports, https://github.com/ruby/ruby/commit/83ecdd1dce18246de631eb3b5d8308145bb269f5 is used.
Among ~455 packages, now 16 packages see build failure. Please check what is going, thank you.
Regards, Mamoru
1. rubygem-actionview https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Same as before.
rubygem-actionview https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui...
Failure: RipperTrackerTest#test_finds_dependencies_with_extra_spaces [test/template/dependency_tracker_test.rb:191]: --- expected +++ actual @@ -1 +1 @@ -["spaces/header", "spaces/form", "messages/message", "events/event", "comments/comment"] +["spaces/header", "spaces/form", "messages/message", "comments/comment"]Discussed on: https://src.fedoraproject.org/rpms/rubygem-actionview/pull-request/4
rubygem-addressable is FIXED: as reported before, this is fixed with https://github.com/ruby/ruby/pull/8813 https://bugs.ruby-lang.org/issues/19992
rubygem-addressable https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Test suite segfaults constantly...
2. rubygem-aruba https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Same as before.
rubygem-aruba https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Cucumber testsuite fails. Some test failures are due to removed irb/ext/save-history, some are due to pry behaving badly? on ruby3.3 (not knowing in details), others I don't know. Already reported https://github.com/cucumber/aruba/issues/910 , no response.
3. rubygem-async_sinatra https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Same as before.
rubygem-async_sinatra https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... `test/test_async.rb:28:in `<main>': uninitialized constant MiniTest (NameError)` Fixing this will perhaps be easy, leaving this to maintainer for now.
rubygem-bootsnap is FIXED: https://github.com/Shopify/bootsnap/pull/460
rubygem-bootsnap https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui...
1) Failure: Bootsnap::KernelRequireTest#test_uses_the_same_duck_type_as_require [/builddir/build/BUILD/bootsnap-1.15.0/usr/share/gems/gems/bootsnap-1.15.0/test/load_path_cache/core_ext/kernel_require_test.rb:26]: Expected #<Process::Status: pid 1413 exit 1> to be success?.I don't know what this means. With ruby 7b8d472100 (around 2023-10-06) test was successful, but with ruby 55c5ebe0a0 (around 2023-10-14) test test fails, not sure what ruby change caused this.
4. rubygem-byebug https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Same as before.
Note that as I posted before, I've already orphaned rubygem-byebug. It seems rubygem-pry-byebug only depends on rubygem-byebug.
rubygem-byebug https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... As discussed before, I don't expect that rubygem-byebug comes to work with ruby3.3. Probably I am going to orphan this.
5. rubygem-childprocess https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Same as before.
rubygem-childprocess https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... `Failure/Error: require 'rubygems/mock_gem_ui'` This file is removed: https://github.com/ruby/ruby/commit/74772840430fc3fca3f5fb0ad585d9cc48f512fb Need to address in childprocess side.
6. NEW: rubygem-liquid https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... ``` 1) Failure: StandardFiltersTest#test_sort_when_property_is_sometimes_missing_puts_nils_last [/builddir/build/BUILD/liquid-4.0.3/usr/share/gems/gems/liquid-4.0.3/test/integration/standard_filter_test.rb:220]: --- expected +++ actual @@ -1 +1 @@ -[{"price"=>1, "handle"=>"gamma"}, {"price"=>2, "handle"=>"epsilon"}, {"price"=>4, "handle"=>"alpha"}, {"handle"=>"delta"}, {"handle"=>"beta"}] +[{"price"=>1, "handle"=>"gamma"}, {"price"=>2, "handle"=>"epsilon"}, {"price"=>4, "handle"=>"alpha"}, {"handle"=>"beta"}, {"handle"=>"delta"}]
2) Failure: StandardFiltersTest#test_sort_natural_when_property_is_sometimes_missing_puts_nils_last [/builddir/build/BUILD/liquid-4.0.3/usr/share/gems/gems/liquid-4.0.3/test/integration/standard_filter_test.rb:248]: --- expected +++ actual @@ -1 +1 @@ -[{"price"=>"1", "handle"=>"gamma"}, {"price"=>2, "handle"=>"epsilon"}, {"price"=>"4", "handle"=>"alpha"}, {"handle"=>"delta"}, {"handle"=>"beta"}] +[{"price"=>"1", "handle"=>"gamma"}, {"price"=>2, "handle"=>"epsilon"}, {"price"=>"4", "handle"=>"alpha"}, {"handle"=>"beta"}, {"handle"=>"delta"}] ```
Actually rawhide also began to fail: https://koschei.fedoraproject.org/package/rubygem-liquid
This is glibc-2.38.9000-20.fc40 qsort behavior change: I posted the issue and proposal on: https://github.com/Shopify/liquid/issues/1759 https://github.com/Shopify/liquid/pull/1760
rubygem-clockwork is FIXED: with backporting upstream PR for mocha and Minitest issue.
rubygem-clockwork https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... `internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb:128:in `require': cannot load such file -- mocha/setup (LoadError)` This mocha issue is already fixed in https://github.com/Rykian/clockwork/pull/64/ . Looks like in addition Minitest issue needs fixing.
rubygem-curb SEEMS FIXED, actually I am not sure what packages' changes fixed this...
rubygem-curb https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui...
Error: test_curlopt_stderr_with_file(TestCurbCurlEasy): Errno::ENOENT: No such file or directory @ rb_sysopen - /tmp/curb_test_curlopt_stderr20231028-3388-1r4xk0 /builddir/build/BUILD/curb-1.0.1/usr/share/gems/gems/curb-1.0.1/tests/tc_curl_easy.rb:30:in `read' /builddir/build/BUILD/curb-1.0.1/usr/share/gems/gems/curb-1.0.1/tests/tc_curl_easy.rb:30:in `test_curlopt_stderr_with_file'Well, successful on x86_64, failing on s390x. But even on s390x this was successful with ruby 7b8d472100 (around 2023-10-06), seeing failure with ruby 55c5ebe0a0 (around 20231014).
rubygem-power_assert is FIXED: As I posted before, rubygem-byebug BR was actually unneeded, so I've removed that dependency.
rubygem-power_assert https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... %check needs BR: rubygem-byebug , which doesn't build. Discussed on: https://github.com/ruby/power_assert/issues/47 , I may try to remove byebug dependency myself.
rubygem-puppet-lint is FIXED: puppet-lint is upgraded to 4.2.1, and rubygem(rspec-collection_matchers) BR is removed.
rubygem-puppet-lint https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui...
No matching package to install: 'rubygem(rspec-collection_matchers) >= 1.0'The above package is already retired.
7. rubygem-railties https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Testsuite is failing, but due to some different reason:
``` Failure: FullStackConsoleTest#test_sandbox [test/application/console_test.rb:133]: "> " expected, but got:
Loading development environment in sandbox (Rails 7.0.8) Any modifications you make will be rolled back on exit â–½. Expected # encoding: ASCII-8BIT # valid: true "Loading development environment in sandbox (Rails 7.0.8)\r\nAny modifications you make will be rolled back on exit\r\n\e[1G\xE2\x96\xBD\e[6n" to include "> ".
rails test test/application/console_test.rb:139
F
Failure: FullStackConsoleTest#test_environment_option_and_irb_option [test/application/console_test.rb:133]: "> " expected, but got:
Loading test environment (Rails 7.0.8) Switch to inspect mode. â–½. Expected # encoding: ASCII-8BIT # valid: true "Loading test environment (Rails 7.0.8)\r\nSwitch to inspect mode.\r\n\e[1G\xE2\x96\xBD\e[6n" to include "> ". ```
rubygem-railties https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui...
Failure: ApplicationTests::AssetsTest#test_precompile_shouldn't_use_the_digests_present_in_manifest.json [test/application/assets_test.rb:299]: Expected "application-c56ef81d122dffa8b257b0546ba1b09bd2d8b97e4aef881de8db9f760b903af6.css" to not be equal to "application-c56ef81d122dffa8b257b0546ba1b09bd2d8b97e4aef881de8db9f760b903af6.css".Not sure what this means.
8. rubygem-ronn-ng https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/pac... Same as before.
rubygem-ronn-ng https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui...
Failure: test_angle_bracket_syntax_HTML(RonnTest): --- /builddir/build/BUILD/ronn-ng-0.9.1/usr/share/gems/gems/ronn-ng-0.9.1/test/angle_bracket_syntax.html 2023-10-28 15:58:23.177074626 +0000 +++ /builddir/build/BUILD/ronn-ng-0.9.1/usr/share/gems/gems/ronn-ng-0.9.1/test/angle_bracket_syntax.html.wrong 2023-10-28 15:58:25.675101802 +0000 @@ -13,5 +13,5 @@ <p>or when <code><WORD></code> is enclosed in backticks.</p> -<p>or when <var>WORD</var> has a <dot.> or <colon>.</colon></dot.></p> +<p>or when <var>WORD</var> has a <dot.> or <foo:colon>.</foo:colon></dot.></p> </div> .libxml2 2.11.5 compatibitily issue, discussed on https://github.com/apjanke/ronn-ng/issues/102 . I think it is enough just to fix test suite as it is.
rubygem-shoulda-matchers is FIXED: With some discussion, this is fixed on ruby side: https://github.com/ruby/ruby/commit/e34e8b93f8fac3ef40ab5ed8672fa003f3b4d9c0 ref: https://github.com/rubygems/rubygems/pull/7128
rubygem-shoulda-matchers https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Lots of:
An error occurred while loading ./spec/unit/shoulda/matchers/action_controller/callback_matcher_spec.rb. Failure/Error: require 'unit_spec_helper' NoMethodError: undefined method `tr' for an instance of PathnameNot sure what this means.
rubygem-stringex is FIXED: Actually the segfault on the testsuite was identified as ruby side bug, and was fixed in:
https://github.com/ruby/ruby/pull/8899 https://bugs.ruby-lang.org/issues/19994
rubygem-stringex https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Also for this one, test suite segfaults constantly...
rubygem-yard is FIXED SO FAR. This is discussed on https://github.com/lsegal/yard/issues/1514 . Although the upstream is still seeking for some different solution, for now I've just removed invalid `yield` usage from the file used in testsuite.
rubygem-yard https://copr.fedorainfracloud.org/coprs/mtasaka/rubygem-newruby-test-3-2/bui... Many messages like:
1) YARD::Handlers::Ruby::YieldHandler only parses yield blocks in methods Failure/Error: raise ProxyMethodError, "Proxy cannot call method ##{meth} on object '#{path}'"Discussed on https://github.com/lsegal/yard/issues/1514 : Most probably this is caused by https://github.com/ruby/ruby/commit/b5e23d3e3b5ff2f5328aa43a2392ebe7c951a222 ruby 3.3.0 now adds more checks for "yield" usage, not knowing how to fix on yard side myself.
Regards, Mamoru