[WIP] upgrading ruby to 3.3

[kbrock]

3.2 ref

Highlights:

• optional: Regexp.timeout = 1.0 or Regexp.new('^a*b?a*()\1$', timeout: Float::INFINITY)
• All methods wishing to delegate keyword arguments through *args must now be marked with ruby2_keywords, with no exception.
• Hash#shift (empty returns nil instead of default)
• Struct no longer needs keyword_init
• removed: Fixnum, Bignum Random::DEFAULT, Struct::Group, Struct::Passwd
• removed: Dir.exists?, File.exists?
• removed: Kernel#=~, Kernel#taint, Kernel#untaint, Kernel#tainted?, Kernel#trust, Kernel#untrust, Kernel#untrusted?
• lib no longer included: libyaml. (was in psych. May need libyaml-dev on ubuntu)
• lib no longer included: libffi. (was in fiddle - only used on windows)
• add Queue#pop, SizedQueue#push, SizedQueue#pop added param timeout:

3.3

Highlights:

punt:

• ERB::Util.html_escape is made faster than CGI.escapeHTML
• no longer need require "set" (supporting older ruby versions)
  • this is not hurting anyone, and this change is not compatible for ruby <3.2 - which we still support

Build Issues

• Tangents
  • update signature for hash_foreach value oVirt/ovirt-engine-sdk-ruby#17
  • Extracted: Update gemspec dependencies for ruby 3 oVirt/ovirt-engine-sdk-ruby#19
  • Lenient call with gvl ruby/ruby#11649
• spec: Support ruby 3.1 and 3.3 error message formats #23203
• Remove Kernel.taint manageiq-smartstate#199
• [WIP] explictly require fiddle more_core_extensions#124

current solution:

gem install --verbose ovirt-engine-sdk -v4.6.0 -- --with-cflags="-Wno-error=incompatible-function-pointer-types -Wno-error=implicit-function-declaration"

[jrafanie]

@kbrock Even if we jump from 3.1 to 3.3 in a release, we should include 3.2 with 3.3 so if they fail in CI, we can determine if it's a problem unique to 3.3 or also a problem in 3.2

[kbrock]

Yea, since we will never use 3.2, I'm mixed on this one.
Already got 2 thumbs up, so I'll add it.
But it seems a waste of cycles and unnecessary granularity

[bdunne]

For build, EL9 ships with ruby 3.0 by default and has dnf modules for 3.1 and 3.3, but not 3.2.

[Fryguy]

so weird that el9 jumped a version 😕

[kbrock]

The gem build failure for ovirt-engine-sdk is frustrating.
I can see that @jrafanie and others had issues building it for oVirt/ovirt-engine-sdk-ruby#11

Hopefully it shouldn't be too hard. Guess this is where it all begins. Wish they had a C guide titled "this is what changed over the past 20 years" - because that code has a bunch of stuff I do not recognize

[jrafanie]

ov_http_request.c:75:29: warning: excess elements in array initializer
   75 |         .reserved = { NULL, NULL }
      |                             ^~~~
ov_http_request.c:75:29: note: (near initialization for
‘ov_http_request_type.function.reserved’)
ov_http_request.c: In function ‘ov_http_request_define’:
ov_http_request.c:347:77: error: ‘rb_cData’ undeclared (first use in this
function)
347 |     ov_http_request_class = rb_define_class_under(ov_module,
"HttpRequest", rb_cData);
|                                                                         
^~~~~~~~
ov_http_request.c:347:77: note: each undeclared identifier is reported only once
for each function it appears in
ov_http_request.c: At top level:
cc1: note: unrecognized command-line option ‘-Wno-self-assign’ may have been
intended to silence earlier diagnostics
cc1: note: unrecognized command-line option ‘-Wno-parentheses-equality’ may have
been intended to silence earlier diagnostics
cc1: note: unrecognized command-line option ‘-Wno-constant-logical-operand’ may
have been intended to silence earlier diagnostics
make: *** [Makefile:248: ov_http_request.o] Error 1

This looks alittle different. There's no option reported in the output to silence it.

[Fryguy]

That's also an error, whereas the others were warnings.

[jrafanie]

That's also an error, whereas the others were warnings.

sorry, it's different than the one we had previously where it told you that you can ignore it by opting into that behavior:

implicit function declarations [-Wimplicit-function-declaration]

oVirt/ovirt-engine-sdk-ruby#11

This one is a undeclared variable if I'm reading correctly.

[micwoj92]

@jrafanie in the linked issue from your last comment you tried to build version 4.6.0 of ovirt engine sdk. This error is because the CI pulls 4.4.1 and it is unrelated. This issue has already been fixed with latest version. See oVirt/ovirt-engine-sdk-ruby@97df88c

[kbrock]

@miq-bot cross_repo_test #23142 including ManageIQ/manageiq-providers-ovirt#676 ManageIQ/manageiq-providers-red_hat_virtualization#54

[Fryguy]

@kbrock I find this site more useful for detailing changes in Ruby:

• 3.2: https://rubyreferences.github.io/rubychanges/3.2.html
• 3.3: https://rubyreferences.github.io/rubychanges/3.3.html

[miq-bot]

This pull request is not mergeable. Please rebase and repush.

[miq-bot]

Checked commit kbrock@2129ca5 with ruby 3.1.5, rubocop 1.56.3, haml-lint 0.51.0, and yamllint
1 file checked, 0 offenses detected
Everything looks fine. 👍