Nicolas Weeger wrote: > Hello. > > Checked bug #1006968 ( > https://sourceforge.net/tracker/index.php?func=detail&aid=1006968&group_id=13833&atid=113833 > ) about GTK client (under Windows) crashing. > > The map itself is ok, server doesn't mind anything, image is correct > (trap is a regular trap => red square). > > But what happens is that, when stepping on the rune, client gets a face > '0' sent. But the 'pixmaps' structure is not initialized (it's > apparently NULL for all items), so client'll crash on gx11.c:989 > (pixmaps[0] == NULL). note that face 0 is never a valid face to display. If memory serves, face 0 is used to denote that there is nothing to display on tha layer anymore. If you look at the images, face 0 is actually a little bug symbol, or at least used to be. > > Checking code, it seems 'pixmaps' is used with 'CONFIG_CACHE', which is > the 'cache images' option in configuration. It's unchecked on my client, > and the 'init_cache_date' function (responsible for loading the pixmaps) > does not get called, thus 'pixmaps[0]' is NULL. > > Tried checking 'cache image', client crash lol. > > So wondering: client crashes because image 0 (question mark) is not > properly handled. Should client check that special case? Should > pixmaps[0] get the question mark anyway? > > Wondering, as i'm not all that familiar with this part of the client. I haven't looked closely at the bug. The real question is then why face is being sent or trying to be displayed. It's unclear to me why the client would be trying to display that different on cached vs non cached image mode. _______________________________________________ crossfire-devel mailing list crossfire-devel at lists.real-time.com https://mailman.real-time.com/mailman/listinfo/crossfire-devel