Mercurial > pidgin.yaz
view doc/imgstore-signals.dox @ 25890:9c0337d4b878
Fix the crash/leaks in statusbox
*** Plucked rev 0cb6e54b (darkrain42@pidgin.im):
Look up the old image before we store the new one in the hash table.
Looking up the new image and then unref:ing it (hence freeing the data
and subsequently returning it) is not good. This was causing problems
with changing buddy icons while using mystatusbox. Refs #9120.
*** Plucked rev d25c9ed7 (darkrain42@pidgin.im):
Don't leak image refs in gtkstatusbox and assertion failure in nullprpl
purple_imgstore_new_from_file gives us a ref and
pidgin_status_box_set_buddy_icon also takes a ref.
author | Paul Aurich <paul@darkrain42.org> |
---|---|
date | Wed, 06 May 2009 18:58:47 +0000 |
parents | e0613cf8c493 |
children |
line wrap: on
line source
/** @page imgstore-signals Image Store Signals @signals @signal image-deleting @endsignals @see imgstore.h <hr> @signaldef image-deleting @signalproto char *(*image_deleting)(const PurpleStoredImage *img); @endsignalproto @signaldesc Emitted when a #PurpleStoredImage is about to be destroyed. This allows for what amounts to weak references. Code can hold onto a pointer to the PurpleStoredImage without actually "holding" a reference. They can then use a signal handler to let them know when their img is about to be destroyed. @param img The image about to be destroyed. @note It's not possible to purple_imgstore_ref() img to save it. @endsignaldef */ // vim: syntax=c.doxygen tw=75 et