On 2023/04/12 16:37, Christian Weisgerber wrote: > Stuart Henderson: > > > Basically it seems that g_debug callers expect to be able to use NULL > > srrings and it would be huge whack-a-mole to clean them up. I wonder > > if it might be better to do something to catch these in g_debug() and > > avoid passing the NULL to vfprintf. > > How would you do this? You would have to process the format string > to see which arguments are pointers and then replace corresponding > NULL arguments. The glib code passes the variadic arguments straight > through to v*printf(). Here's how that looks: > > g_debug() macro expands to > -> g_log() calls > -> g_logv() calls > -> g_strdup_vprintf() calls > -> vasprintf() > or > -> _g_vsnprintf() macro expands to > -> vsnprintf() > or > -> g_log_structured() calls > -> g_strdup_vprintf() calls > -> vasprintf() > or > -> _g_vsnprintf() macro expands to > -> vsnprintf()
Perhaps by calling a different function instead of v*printf that does most of the same things except not syslogging for null %s prints..
