Branch: refs/heads/blead
Home: https://github.com/Perl/perl5
Commit: 3be5b33e30b54e62002abd6d1e772e3ee110e89e
https://github.com/Perl/perl5/commit/3be5b33e30b54e62002abd6d1e772e3ee110e89e
Author: Karl Williamson
Date: 2022-04-08 (Fri, 08 Apr 2022)
Changed paths:
M dist/threads/t/stack.t
M dist/threads/t/stack_env.t
Log Message:
---
threads/t/stack*.t: Add is() fcn
These tests roll their own ok() functions, which is ok until there is a
failure, and you don't know what the failing value is. So add an is()
wrapper to them that does return the expected and actual values.
Commit: 601bde632530766686cd3bd4ce0b3a823b87e53d
https://github.com/Perl/perl5/commit/601bde632530766686cd3bd4ce0b3a823b87e53d
Author: Karl Williamson
Date: 2022-04-08 (Fri, 08 Apr 2022)
Changed paths:
M dist/threads/t/stack.t
M dist/threads/t/stack_env.t
Log Message:
---
/threads/t/stack*.t: Use mnemonics
Use mnemonics instead of hard-coded numbers throughout.
Commit: 31898f8b00349cec9507fc23a1899d8e2e457ca7
https://github.com/Perl/perl5/commit/31898f8b00349cec9507fc23a1899d8e2e457ca7
Author: Karl Williamson
Date: 2022-04-08 (Fri, 08 Apr 2022)
Changed paths:
M dist/threads/t/stack.t
M dist/threads/t/stack_env.t
Log Message:
---
threads/t/stack* Generalize larger min stack sizes
This changes the tests so that they will work on systems that require
larger minimum stack sizes than what we initially try. The initial test
is skipped in this case, but the rest of the tests are adjusted from the
new information.
This was the least disruptive way I could come up with that didn't
assume things worked correctly before running the tests, yet still
adapted to a platform that has a larger minimum requirement than we
guess.
One could increase the initial guess to the maximum of any box found so
far (this would mean doubling it in this case), but what happens if a
larger case comes around?
Commit: 0eeeceed10b2ab27746282e8cfb05afb1549d925
https://github.com/Perl/perl5/commit/0eeeceed10b2ab27746282e8cfb05afb1549d925
Author: Karl Williamson
Date: 2022-04-08 (Fri, 08 Apr 2022)
Changed paths:
M dist/threads/t/stack.t
M dist/threads/t/stack_env.t
Log Message:
---
threads/t/stack*: Add comment
I noticed that the first real test in each of these could succeed by
happenstance even if the implementation is broken. If the numbers we
use happen to be the same as the system default, we'll get true even if
the system ignored the request to use our numbers.
Fixing the test is a chicken and egg problem, as to find out what the
default value is requires using the functions the test files are trying
to test. A unit could be added to Configure, but that seems too much
work for just this purpose.
The comments added by this commit suggest using an out-of-the-ordinary
value instead of the power of 2 one that is currently used. That would
minimize the likelihood of happenstance.
Compare: https://github.com/Perl/perl5/compare/df12890b5e8c...0eeeceed10b2