mauke
mauke
> they build for all listed architectures simultaneously -- there is only one Configure run for ALL of them combined, not one per architecture The only way that could work...
> This does not look like a copy vs reference issue. I disagree. Modifying `@_` does not alter the caller's argument list. Only the elements of `@_` (i.e. `$_[0]`, `$_[1]`,...
No repro here: ``` $ perl bug.pl [abc 17:38:42] [абв 17:38:42] ``` But that's with `PERL_UNICODE=SAL` in the environment. Without that, the results I'd expect to see are: ``` $...
Ah. I can reproduce it with the original code using `LC_TIME=C perl bug.pl`, so the issue is definitely locale-related.
Bisecting using `LC_TIME=C ../perl5/Porting/bisect.pl --start=v5.38.0 --end=v5.40.0 -e 'use POSIX qw(strftime); strftime("\x{100}", localtime) ne "" or die'` leads to commit 492c28719682673e6c19378edf6824d79784b33e. @khwilliamson ? ``` bad - non-zero exit from ./perl -Ilib...
> Only ASCII characters are legal in the C locale. You are calling `strftime` with illegal inputs. Do you have a reference for that? Because I tried it in C...
It's not from this development cycle. The bug report was against 5.40.2 and the bisected commit was already in v5.39.2.
Isn't this just [pointed](https://hackage.haskell.org/package/pointed)?
Similar, but with a lot less casting: ```diff diff --git hv.c hv.c index 6fea67d7af..ec7df5a8bb 100644 --- hv.c +++ hv.c @@ -177,6 +177,24 @@ S_new_he(pTHX) #endif +struct shared_he { + struct...