Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-18 Thread Paul Eggert
On 2024-06-18 06:02, Bruno Haible wrote: tz=NULL means universal time (a.k.a. GMT) in our strftime.h, but is undefined behaviour for NetBSD's functions. Thanks for the fix. I tweaked it via the attached. The "0" at the end of "UTC0" is better if the zoneinfo files are missing.From 562ed134d6cf

Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-18 Thread Bruno Haible
Paul Eggert wrote: > > After the Solaris crash fix, the nstrftime tests still fail, due to this > > output: > > > > <-00>0: expected "1970-01-01 00:00:00 - (-00)", got "1970-01-01 > > 00:00:00 + (-00)" > > <-00>0: expected "1985-11-05 00:53:21 - (-00)", got "1985-11-05 > > 00:53:21 +

Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-17 Thread Paul Eggert
On 2024-06-17 07:51, Bruno Haible wrote: After the Solaris crash fix, the nstrftime tests still fail, due to this output: <-00>0: expected "1970-01-01 00:00:00 - (-00)", got "1970-01-01 00:00:00 + (-00)" <-00>0: expected "1985-11-05 00:53:21 - (-00)", got "1985-11-05 00:53:21 +

Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-17 Thread Bruno Haible
Paul Eggert wrote: > Thanks for reporting and fixing that. For NetBSD, a better fix is to use > the native timezone_t, so that Gnulib doesn't need to set and later > restore TZ. Interesting. So, we have at least one platform now, where nstrftime is multithread-safe. Only an indentation nit: 2

Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-17 Thread Paul Eggert
Thanks for reporting and fixing that. For NetBSD, a better fix is to use the native timezone_t, so that Gnulib doesn't need to set and later restore TZ. I attempted to do by installing the attached.From 17432773bb157fe9bd11137edd42cd7be35e1dea Mon Sep 17 00:00:00 2001 From: Paul Eggert Date: Mon

Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-17 Thread Bruno Haible
After the Solaris crash fix, the nstrftime tests still fail, due to this output: <-00>0: expected "1970-01-01 00:00:00 - (-00)", got "1970-01-01 00:00:00 + (-00)" <-00>0: expected "1985-11-05 00:53:21 - (-00)", got "1985-11-05 00:53:21 + (-00)" <-00>0: expected "2001-09-09 01:46:

Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-17 Thread Bruno Haible
Paul Eggert did: > (__strftime_internal): Simplify calculation of zone to be closer > to what glibc does. This causes 4 test suite failures on Solaris 11.4, from a null pointer reference: FAIL: test-c-nstrftime-1.sh === ../../gltests/test-c-nstrftime-1.sh: line 3: 9936: M

Re: [PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-17 Thread Bruno Haible
Paul Eggert did: > (set_tz, revert_tz): Declare, as they’re now extern. This causes link errors on NetBSD 10 (caught by the CI): gcc -Wno-error -g -O2 -L/home/bruno/lib -L/usr/pkg/lib -o test-c-nstrftime test-c-nstrftime.o libtests.a ../gllib/libgnu.a libtests.a ../gllib/libgnu.a libtests.a

[PATCH 6/7] nstrftime, time_rz: don’t depend on tzname

2024-06-16 Thread Paul Eggert
* lib/strftime.c (HAVE_TZNAME_ARRAY) [_LIBC]: Remove. All uses removed. (__strftime_internal): Simplify calculation of zone to be closer to what glibc does. However, in Gnulib do not look at tzname as this is problematic for all the usual reasons; instead, use the underlying strftime with %Z, maki