This is the mail archive of the
mailing list for the glibc project.
Re: [PATCH 1/2] Merge mktime, timegm from upstream Gnulib
- From: Paul Eggert <eggert at cs dot ucla dot edu>
- To: Florian Weimer <fweimer at redhat dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Fri, 9 Nov 2018 19:51:31 -0800
- Subject: Re: [PATCH 1/2] Merge mktime, timegm from upstream Gnulib
- References: <email@example.com> <firstname.lastname@example.org> <email@example.com>
Florian Weimer wrote:
It happens for the /usr/share/zoneinfo/America/Sao_Paulo file from
tzdata-2018e-1.fc27. So the test case is not properly isolated, either.
The test case should work regardless of what's in that file, and should work
even if that file does not exist or if there's an I/O error reading it. As I
vaguely recall, the original bug report was against TZ=America/Sao_Paulo and
that's why the test case uses that particular setting.
It'd probably be good to add "<-03>3<-02>,M11.1.0/0,M2.3.0/0" to the tz_strings
array in that test case, to do a southern-hemisphere-style test even if the
Sao_Paulo file contains garbage or does not exist. I don't recall the original
problem well enough to know whether such an addition to tz_strings would suss
out the Sao_Paulo bug.