Bug reports #4031, #4914, and #4927 relate that Layers & Channel layers become "sticky" and one layer retains focus and gets drawing edits despite what the visual cues and highlighting in the dialog box suggest. I've uploaded a patch to CVS (ChangeLog CVS 1.2077) that corrects one generator of this problem, but suspect there may be others. Suspicions center on a flag defined locally in layers_dialog.c, suspend_gimage_notify. If this flag becomes positive, it supresses the actual change of layers in the GimpImage structure, engendering conditions cited in the bug reports. See details at http://idt.net/~gosgood/gimp-patch/patch04.html and the respective bug reports. The flag appears to be used to consolidate multiple display flushes into one transition of the active layer, minimizing display thrashing. Could anyone with L&C dialog expertise also look at suspend_gimage_notify to see if it is being used in an entirely sane manner? The flag should be normally zero; it becomes non-zero after an initial call to gdisplay_flush(), but should go back to zero so long as signal handling remains symmetrical (Is it?). Also, after getting a current CVS version, could the bug report contributors see if the problem happily goes away? Thank you for your help Garry Osgood
