Bug #15736: r67361 Update to ruby/spec@e81b3cd - spec failures on MinGW ?

• Status: Closed
• Assignee: -
• Target version: -
• ruby -v: ruby 2.7.0dev (2019-03-29 trunk 67365) [x64-mingw32]
• Backport: 2.4: UNKNOWN, 2.5: UNKNOWN, 2.6: UNKNOWN

[MSP-Greg (Greg L)]

ruby-loco (MinGW trunk) had the following failures in spec tests as of r67365 (r67357 passed)

1)
The --enable and --disable flags can be used with gems FAILED
Expected "nil"
 to equal "\"constant\""

/ruby/spec/ruby/command_line/feature_spec.rb:6:in `block (2 levels) in <top (required)>'
/ruby/spec/ruby/command_line/feature_spec.rb:3:in `<top (required)>'

2)
The --enable and --disable flags can be used with gem FAILED
Expected "nil"
 to equal "\"constant\""

/ruby/spec/ruby/command_line/feature_spec.rb:13:in `block (2 levels) in <top (required)>'
/ruby/spec/ruby/command_line/feature_spec.rb:3:in `<top (required)>'

3)
The --enable and --disable flags can be used with did_you_mean FAILED
Expected "nil"
 to equal "\"constant\""

/ruby/spec/ruby/command_line/feature_spec.rb:20:in `block (2 levels) in <top (required)>'
/ruby/spec/ruby/command_line/feature_spec.rb:3:in `<top (required)>'

[Eregon (Benoit Daloze)]

@MSP-Greg (Greg L) Could you investigate why they fail?
Those specs look fine and portable from a quick look.

Does "The --enable and --disable flags can be used with frozen-string-literal" passes?
That's very similar, could it be a bug of command-line parsing or features handling on MRI on Windows?

[MSP-Greg (Greg L)]

@Eregon (Benoit Daloze)

Could you investigate why they fail?

Yes. I haven't had time to check it locally, will check tonite...

[MSP-Greg (Greg L)]

@Eregon (Benoit Daloze)

Sorry & fixed. My test code had some disable=gems in test-all, and I never cleared it for spec. The new tests aren't happy with that.

Note that I run tests in ruby-loco from the install folder, so I don't use make. Essentially the same as in ruby/spec...

Feel free to close.

[Eregon (Benoit Daloze)]

MSP-Greg (Greg L) wrote:

Sorry & fixed. My test code had some disable=gems in test-all, and I never cleared it for spec. The new tests aren't happy with that.

Great.
Now, it is weird that the command line doesn't override the environment variable (I assume you use RUBYOPT):

$ RUBYOPT=--disable-gems ruby --enable=gems -e 'p defined?(Gem)'
nil

That seems a bug.

$ ruby --disable=gems --enable=gems -e 'p defined?(Gem)'     
"constant"
$ ruby --enable=gems --disable=gems -e 'p defined?(Gem)'
nil

works as expected.

[Eregon (Benoit Daloze)]

• Related to Bug #15738: RUBYOPT disables have precedence over the command line added

[Eregon (Benoit Daloze)]

• Status changed from Open to Closed

I'll close this issue, I filed #15738 for the strange behavior.