Front page | perl.perl5.porters |
Postings from October 2014
On Thu Oct 30 08:47:32 2014, Hugmeir wrote:
> I was trying to build perl on Android and spotted this failure:
>
> t/../ext/XS-APItest/t/svpeek ... 1..52
> ok 1 - use XS::APItest;
> ok 2 - $/
> ok 3 - $\
> ok 4 - $.
> ok 5 - $,
> ok 6 - $;
> ok 7 - $"
> ok 8 - $:
> ok 9 - $~
> ok 10 - $^
> ok 11 - $=
> ok 12 - $-
> ok 13 - $!
> ok 14 - $?
> ok 15 - $|
> not ok 16 - $1
> # Failed test ' $1'
> # at t/svpeek.t line 31.
> # got: 'PVMG("t"\0)'
> # expected: 'PVMG()'
> # Looks like you failed 1 test of 52.
> FAILED at test 16
> Failed 1 test out of 1, 0.00% okay.
> ../ext/XS-APItest/t/svpeek.t
>
> The test itself is this:
>
> "abc" =~ m/(b)/; # Don't know why these magic vars have this content
> is (DPeek ($1), 'PVMG()', ' $1');
>
> where DPeek is a wrapper around sv_peek() from dump.c. I may be
> missing something, but it looks like the test itself has two wrongs:
> First, it needs to use $1 for the value of the previous match to get
> populated; second, if it wanted 'PVMG()', then the regex prior should
> either not capture anything, or fail.
>
> The attached patch solves the issue for me, but I'm rather curious as
> to why this only affected Android.
That test was last touched in 07308ed1, a recent Test-Simple upgrade. At the time I saw it, but assumed the test was just a little too sensitive to internals, or that using like() on $1 didn’t work any more.
This seems to show the difference:
$ ./perl -Ilib -MXS::APItest= -e ' "abc" =~ m/(b)/;; warn DPeek($1)'
PVMG() at -e line 1.
$ ./perl -Ilib -MXS::APItest= -e ' "abc" =~ m/(b)/;; "$1"; warn DPeek($1)'
PVMG("b"\0) at -e line 1.
So, somewhere, something is stringifying the $1 (or calling get-magic some other way).
--
Father Chrysostomos
---
via perlbug: queue: perl5 status: new
https://rt.perl.org/Ticket/Display.html?id=123089
Thread Next