lib/util:tests/time.c - "test_timestring" - fix it on Solaris

[Jelmer Vernooij]

Hi Matthias,

On Wed, 2010-12-22 at 10:13 +0100, Matthias Dieter Wallnöfer wrote:
> another patch proposal. This simply specifies more precisely the output 
> of the date and it works on both Solaris and Linux (and I bet on other 
> Unix systems as well). In my eyes this should be definitely okay for 
> passing "samba4.local.time".
That seems reasonable to me. Have you checked how strict the callers of
timestring() rely on its format?

Cheers,

Jelmer

> > diff --git a/lib/util/time.c b/lib/util/time.c
> > index a1f2b08..770ebc4 100644
> > --- a/lib/util/time.c
> > +++ b/lib/util/time.c
> > @@ -387,7 +387,7 @@ _PUBLIC_ char *timestring(TALLOC_CTX *mem_ctx, 
> > time_t t)
> >         /* Some versions of gcc complain about using some special format
> >          * specifiers. This is a bug in gcc, not a bug in this code. See a
> >          * recent strftime() manual page for details. */
> > -       strftime(tempTime,sizeof(tempTime)-1,"%c %Z",tm);
> > +       strftime(tempTime,sizeof(tempTime)-1,"%a %b %e %X %Y %Z",tm);
> >         TimeBuf = talloc_strdup(mem_ctx, tempTime);
> >  #else
> >         TimeBuf = talloc_strdup(mem_ctx, asctime(tm));
> Cheers,
> Matthias
> 
> Matthias Dieter Wallnöfer wrote:
> > Okay, I thought such a little difference (the "0" or not) wouldn't 
> > matter - but I will revert.
> >
> > Matthias
> >
> > Jelmer Vernooij wrote:
> >> On Mon, 2010-12-06 at 11:38 +0300, Matthieu Patou wrote:
> >>> I guess you have to introduce a test in  lib/util/wscript_configure to
> >>> define
> >>> HAVE_CORRECT_STRFTIME by testing the output of strftime() and if the
> >>> output is not ok then we should use asctime (at least on sun8 it 
> >>> seems ok).
> >> I think we should rather try to keep all platform-specific workarounds
> >> in lib/replace. So if the strftime() doesn't do what we expect it to I
> >> think we should add a strftime() replacement to lib/replace.
> >>
> >> Cheers,
> >>
> >> Jelmer
> >>
> >>> Have a look at changeset 4ea7d4694a8353fc55ecd12cb09b9c91ffde7b3f
> >>> On 06/12/2010 05:59, Andrew Bartlett wrote:
> >>>
> >>> Matthieu.
> >>>> On Sun, 2010-12-05 at 23:10 +0100, Matthias Dieter Wallnöfer wrote:
> >>>>> The branch, master has been updated
> >>>>>          via  654e010 lib/util:tests/time.c - "test_timestring" - 
> >>>>> fix it on Solaris
> >>>>>         from  55dba7b s4:cluster/cluster.h - fix another gcc 3.4 
> >>>>> "struct" warning
> >>>>>
> >>>>> http://gitweb.samba.org/?p=samba.git;a=shortlog;h=master
> >>>>>
> >>>>>
> >>>>> - Log 
> >>>>> -----------------------------------------------------------------
> >>>>> commit 654e0102ddb0acaaf45fb55c15818722235fcc9f
> >>>>> Author: Matthias Dieter Wallnöfer<mdw at samba.org>
> >>>>> Date:   Sun Dec 5 22:20:06 2010 +0100
> >>>>>
> >>>>>       lib/util:tests/time.c - "test_timestring" - fix it on Solaris
> >>>>>
> >>>>>       Solaris returns "Thu Jan 01" and not "Thu Jan  1" - 
> >>>>> therefore proof for
> >>>>>       both.
> >>>> I'm not sure that this patch is correct.
> >>>>
> >>>> Given we had a test for a particular value, and we carefully build the
> >>>> value with strftime() if available, why does this patch accept 
> >>>> different
> >>>> values, rather than fix the the timestring() function?
> >>>>
> >>>> In general, I'm suspicious of patches that fix tests by changing the
> >>>> test result rather than fixing the original function.
> >>>>
> >>>> Perhaps none of the callers care, as long as it's clear to human 
> >>>> readers
> >>>> of the string, but there are rather a lot of callers to this function.
> >>>>
> >>>> Andrew Bartlett
> >>>>
> >>
> >
> >
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20101222/3c7f5a85/attachment.pgp>