> > Looking at the code, I see that the folder.changed event is passed an > EMEventTargetFolder structure, which is contains a flag > EM_EVENT_FOLDER_NEWMAIL (although curiously this flag is inverted in > em_event_target_new_folder() - not > sure why). > > Therefore the plugin code in new-mail-notify.c could potentially test this > flag to decide wether to send the dbus message - this would save having to > code a brand new plugin event type. >
Looking further, I see it is worse than this. In function real_flush_updates() (which originates the folder.changed event) the up->new field is used to set flag EM_EVENT_FOLDER_NEWMAIL. However, up->new is not an indication of truely new mail, but rather is a count of mail added to this folder (regardless of its unread status). So this is why even things like 'play sound' are incorrect - since this same function triggers that notification if up->new is non-zero. I'll keep digging ... ----------------------------------------- Email sent from www.ntlworld.com Virus-checked using McAfee(R) Software Visit www.ntlworld.com/security for more information _______________________________________________ Evolution-hackers mailing list [email protected] http://mail.gnome.org/mailman/listinfo/evolution-hackers
