t/run/locale.t: Test failures on OpenBSD-6.4 unthreaded builds #16745
Comments
From @jkeenanHere is a really weird one: On OpenBSD-6.4, the last two unit tests in I first noticed this in smoke-test reports submitted by Carlos Guevara. ##### ... Test failures: Note that t/run/locale.t only failed on unthreaded builds. We don't Over the last week I modified t/run/locale.t to properly account for Also over the last week I have been able to set up an OpenBSD-6.4 VM ##### $ ./perl -Ilib -V $ cd t; ./perl harness -v run/locale.t; cd - Next, an *unthreaded* build of blead: ##### [perl -V output attached] $ cd t; ./perl harness -v run/locale.t; cd - Finally, this smoke-test report shows this pattern of failures: ##### Applied patches: Skipped tests: Test failures: (Ignore the failures in cpan/Sys-Syslog/t/syslog.t. We're seeing them This perhaps can be seen more clearly like this: ##### Any ideas? Thank you very much. |
From @jkeenanSummary of my perl5 (revision 5 version 29 subversion 5) configuration: Characteristics of this binary (from libperl): |
From @sisyphusOn Sat, 10 Nov 2018 17:02:31 -0800, jkeenan@pobox.com wrote:
This issue also cropped up a week ago at: Cheers, |
|
The RT System itself - Status changed from 'new' to 'open' |
From @khwilliamsonOn 11/10/18 6:02 PM, James E Keenan wrote:
See my response on monks
|
From @khwilliamsonOn 11/10/18 6:02 PM, James E Keenan (via RT) wrote:
Run the attached script on a DEBUGGING build with runtime switch -DLv |
From @khwilliamson |
From @jkeenanOn Sun, 11 Nov 2018 18:00:27 GMT, public@khwilliamson.com wrote:
I had to add '-MPOSIX' to the command-line invocation. See attached. Thank you very much. |
From @jkeenan[perl] 7 $ ./perl -Ilib -V:config_args [perl] 8 $ ./perl -Ilib -DLv /home/jkeenan/learn/perl/133649.pl EXECUTING... Undefined subroutine &POSIX::setlocale called at /home/jkeenan/learn/perl/133649.pl line 3. [perl] 9 $ ./perl -Ilib -MPOSIX -DLv /home/jkeenan/learn/perl/133649.pl EXECUTING... locale.c:2203: setlocale(LC_ALL, "LC_COLLATE=foo_BAR;LC_CTYPE=foo_BAR;LC_MESSAGES=foo_BAR;LC_MONETARY=foo_BAR;LC_NUMERIC=foo_BAR") returned "LC_COLLATE=foo_BAR;LC_CTYPE=foo_BAR;LC_MESSAGES=foo_BAR;LC_MONETARY=foo_BAR;LC_NUMERIC=foo_BAR" |
From @jkeenanOn Sun, 11 Nov 2018 18:51:32 GMT, jkeenan wrote:
Note: I tried switching the 2 lines of code around. I also tried with 'C' instead of 'C.UTF-8'. I got the same "Can't fix broken locale name" error in all cases. -- |
From @khwilliamsonOn 11/11/18 11:58 AM, James E Keenan via RT wrote:
[snip]
Try compiling the attached C program, and run it with a single argument, like ./a.out en_US.UTF-8 and post the results |
From @khwilliamson#include <stdio.h>
#include <locale.h>
#include <stdlib.h>
int main(int argc, char * argv[]) {
char * result;
fprintf(stdout, "setlocale(LC_ALL, \"C\") returns ");
result = setlocale(LC_ALL, "C");
if (result) {
fprintf(stdout, "'%s'\n", result);
}
else {
fprintf(stdout, "NULL\n");
}
fprintf(stdout, "setlocale(LC_CTYPE, \"%s\") returns ", argv[1]);
result = setlocale(LC_CTYPE, argv[1]);
if (result) {
fprintf(stdout, "'%s'\n", result);
}
else {
fprintf(stdout, "NULL\n");
}
fprintf(stdout, "setlocale(LC_ALL, NULL) returns ");
result = setlocale(LC_ALL, NULL);
if (result) {
fprintf(stdout, "'%s'\n", result);
}
else {
fprintf(stdout, "NULL\n");
}
exit(0);
} |
From @jkeenanOn Sun, 11 Nov 2018 22:01:56 GMT, public@khwilliamson.com wrote:
Here's a sampler: ##### Thank you very much. -- |
From @jkeenanOn Sun, 11 Nov 2018 22:59:57 GMT, jkeenan wrote:
I just realized I failed to include '-8' in some of those names. So here's a different sample: ##### $ for l in `cat locales_some.txt`; do ./khw-testlocale-20181111 $l;done setlocale(LC_ALL, "C") returns 'C' |
From @jkeenanOn Sun, 11 Nov 2018 01:02:31 GMT, jkeenan@pobox.com wrote:
An additional instance of this from the smoker: http://perl5.test-smoke.org/report/73918 -- |
From @khwilliamsonOn 11/11/18 4:08 PM, James E Keenan via RT wrote:
It looks like this is a result of this OS using a different syntax for |
From @khwilliamsonOn 11/11/18 4:12 PM, James E Keenan via RT wrote:
I'm wondering why 6.3 worked, but not 6.4. I have a suspicion. Please |
From @jkeenanOn Mon, 12 Nov 2018 16:48:57 GMT, public@khwilliamson.com wrote:
Unfortunately, our openbsd-6.3 smoker met an early demise and it would take a fair amount of work to revive it. Attaching the config.h for openbsd-6.4 (from the nonthreaded DEBUGGING build previously mentioned). Thank you very much. |
From @jkeenanOn Mon, 12 Nov 2018 17:49:53 GMT, jkeenan wrote:
[snip]
Okay, I managed to revive that OpenBSD-6.3 VM. Please find 3 files attached. But note ... Although this VM reports having the same number of locales available as our OpenBSD-6.4 VM, only 2 of them -- C and POSIX -- were found while running t/run/locale.t. I don't know enough about this OS to speculate as to the reason why. As a consequence, only 2 unit tests in that file were run, so I can't meaningfully compare the results with those we're getting on OpenBSD-6.4. Also note that there were no significant differences between the config.h files generated by building blead on the two OS versions. Thank you very much. -- |
From @jkeenanC |
From @jkeenan# locales available: C POSIX |
From @jkeenanOn Mon, 12 Nov 2018 16:48:57 GMT, public@khwilliamson.com wrote:
Karl, Your commit 9f77c4a in your current smoke-me/khw-locale branch mitigates the problem. In the following two smoke test reports there are no failures in t/run/locale.t on OpenBSD-6.4: http://perl5.test-smoke.org/report/74334 Feel free to merge. Thank you very much. |
From @khwilliamsonThis has been fixed in blead by Commit e1895ad |
|
@khwilliamson - Status changed from 'open' to 'pending release' |
From andrew@afresh1.comOn Mon, Nov 19, 2018 at 02:29:05PM -0800, Karl Williamson via RT wrote:
I applied this to a 5.28 release (minus the perldelta changes) and still |
From @jkeenanOn 11/25/18 8:04 PM, Andrew Fresh via RT wrote:
Can you post either a smoke-test result or report of failure plus perl -V? Also, ISTR your mentioning that the way you build the perl package for jimk |
From andrew@afresh1.comOn Sun, Nov 25, 2018 at 08:47:38PM -0500, James E Keenan wrote:
It seems I did something wrong the last time, not sure what yet, but l8rZ, I wish life had an UNDO function. |
From @khwilliamsonThank you for filing this report. You have helped make Perl better. With the release today of Perl 5.30.0, this and 160 other issues have been Perl 5.30.0 may be downloaded via: If you find that the problem persists, feel free to reopen this ticket. |
|
@khwilliamson - Status changed from 'pending release' to 'resolved' |
Migrated from rt.perl.org#133649 (status was 'resolved')
Searchable as RT133649$
The text was updated successfully, but these errors were encountered: