in reply to Re^3: Is it ever legitimate to override $^O ? -- Cwd
in thread Is it ever legitimate to override $^O ?

the following Cwd version acts in the opposite way

Heh ... that's pretty flakey, too. (I see the same.)

What I'd like to find is some code where the behaviour changes between perl-5.32 and perl-5.34.
Apart from t/01-basic.t, everything I've tested (including your fakeOS.pl) is doing the same thing on both of those versions of perl.

If I take your fakeOS.pl, change $^O = 'Unix'; to $^O = 'MSWin32' and run it on Ubuntu-20.04, I get:
$ perl fakeOS.pl Use of uninitialized value $abs_path in concatenation (.) or string at + fakeOS.pl line 10.
and
$ perl -MTest::Harness fakeOS.pl /home/sisyphus/pscrpt
Again, however, it's the same behaviour on perl-5.32 and perl-5.34.
Interestingly, that error message when run without loading Test::Harness hints at being perhaps connected to the failure that t/01-basic.t is triggering on perl-5.34 and later.

Dunno ... I think I just need to find the time to make one or two bug reports and let the experts sort it out.

Cheers,
Rob

Replies are listed 'Best First'.
Re^5: Is it ever legitimate to override $^O ? -- Cwd
by syphilis (Archbishop) on Aug 04, 2022 at 13:02 UTC