Re: Patch for scm_strftime() in stime.c

[Andreas Vögele]

Andreas Vögele wrote:

> Hi,
>
> on HP-UX, test-suite/time.test fails. The problem is that there may be 
> a mismatch between the time zone name and the daylight savings time 
> setting.

I thought about this problem. It's probably better to fix the test 
instead of scm_strftime(). The test gets the local time and then sets 
the time zone to "ZOW". The daylights savings time indicator, which is 
currently set in Central Europe, is not unset. Since "ZOW" is not known 
to be a time zone where dst is in effect the function strftime() 
outputs an empty  time zone name on HP-UX. BTW, that explains why the 
time test didn't fail a couple of weeks ago. At that time dst wasn't in 
effect in Central Europe.

The patch that I sent this afternoon uses mktime() in scm_strftime() to 
fix the tm_isdst field. But IMHO it's better to fix the test even if 
that means that HP-UX and glibc-based systems behave differently in 
some cases. Or do you think that scm_strftime() should silently fix 
mismatches between the time zone name and dst?

If yes, then I'd remove #idef __hpux since other systems might behave 
similarly to HP-UX. On HP-UX, mktime() doesn't fail if the time zone 
and dst mismatch. Instead, tm_isdst is modified. I don't know about 
other systems, though.

A patch for time.test is attached.

time.test.diff
Description: Binary data