This is exactly what the [EMAIL PROTECTED] hack is supposed to fix. I'll contact the Glib maintainer again about including support for emitting "(null)" when a null pointer is passed for a %s printf argument.
Tell me more about your Solaris. What revision is it? Which CPU type? Aaron On Sat, 2006-09-16 at 09:24 +0200, Lars Kneschke wrote: > Paul J Stevens <[EMAIL PROTECTED]> schrieb: > >Lars Kneschke wrote: > >> Hello! > >> > >> Today was debuging day. I tried to solve the random crashes. I found one > >> problem in glib2. After moving around some ifdef's i solved one problem. > > > >Tell use more, please! > After configure is run, i apply following patch: --- config.h.orig Fr > Sep 15 15:08:55 2006 > +++ config.h Fr Sep 15 15:10:20 2006 > @@ -148,7 +148,7 @@ > #define HAVE_GETTEXT 1 > > /* define to use system printf */ > -/* #undef HAVE_GOOD_PRINTF */ > +#define HAVE_GOOD_PRINTF 1 > > /* define to support printing 64-bit integers with format I64 */ > /* #undef HAVE_INT64_AND_I64 */ > @@ -337,7 +337,7 @@ > #define HAVE_VALUES_H 1 > > /* Define to 1 if you have the `vasprintf' function. */ > -#define HAVE_VASPRINTF 1 > +/* #define HAVE_VASPRINTF 1 */ > > /* Define to 1 if you have the `vprintf' function. */ > #define HAVE_VPRINTF 1 > I just shuffle around some defines.The code which requires this defines is > in glib/gprintf.c about line ~300(function gint g_vasprintf). With the > standard config.h the first if clause get's used. With my modified config.h > the last if clause get's used. And now i don't get any stack corruptions any > more.I have no idea why. I was just poking around with the help of my > brother. :-) > > > > _______________________________________________ > Dbmail-dev mailing list > Dbmail-dev@dbmail.org > http://twister.fastxs.net/mailman/listinfo/dbmail-dev
