> -----Original Message-----
> From: Vibhu Mohindra [mailto:[EMAIL PROTECTED]]
>
> This is with respect to the CVS version of Kannel. The SMPP driver's
> calculation of the validity and deferred fields for messages
> to the SMSC
> is incorrect. (I noticed this because it worked with my simulator at
> 2340 hrs but not at 0100 hrs. Sure enough in the first case
> it correctly
> said I was +6 UTC, but in the second it said I was -18 UTC. Of course
> the simulator then returned a NAK).
> localtime() originated from SysV. Upon return it sets a global called
> timezone, which stores the offset from UTC. This isn't nice for
> multithreaded programs. (Though in this particular case it's probably
> alright since one's UTC offset won't change from one call to
> the next). A
> POSIX.1c variant is localtime_r(). This gives you the offset
> required in
> a local variable (in addition to setting the global timezone). While a
> solution with localtime() would be fine, the only drawback is
> it doesn't
> take into account daylight savings which localtime_r() does.
I don't understand what your patch is doing. my manual pages clearly
claim :
<snip>
The localtime() function converts the calendar time timep to broken-time
representation ...
The localtime_r () function does the same, but stores the data in a
user-supplied struct. It need not set tzname.
</snip>
So using localtime_r() over localtime() has no advantage to us
(especially as we use a serializing wrapper over the system's
localtime() ), except making the code less portable as localtime_r() is
not available on all platforms. the struture member you are using,
tm_gmtoff, is returned by localtime() too, as both are returning the
same structure, and is only available on glibc (libc6 and above) making
the code even less protable!
I agree that the current offset calculation is not correct for certain
edge conditions, but we need to fix it by fixing the algorithm, not by
introducing portability issues.
I vote -1 on that patch.
--
Oded Arbel
m-Wise mobile solutions
[EMAIL PROTECTED]
+972-9-9581711 (116)
+972-67-340014
::..
If you lie to the compiler, it will get its revenge.
-- Henry Spencer
>
> My changes are attached. I updated my copy of the CVS tree at
> approximately 17:30 GMT on 21 Aug and this is a diff to it.
>
> Vibhu
>
