Andy> There were several reasons why redisplay optimisations weren't Andy> kicking in for gutters. This should fix the worst two offenders Andy> - extent changes in the buffer and having more than one gutter Andy> on a frame. This should cause the general flickering people have Andy> been seeing. I would be interested to know if it doesn't I still get annoying flickering on each mouse event if I do M-x occur followed by moving the mouse around in the *Occur* buffer. Did I mention you could get access to my binary to try this yourself...?

Andy> Annoying flickering of what - the gutter? I'll try and investigate. I'd Yes, the gutter. It is redrawn on every mouse event in the *Occur* buffer (even horizontal ones that don't change any extents). My default face is grey on black. I see the entire gutter drawn with a greyish background before being redrawn with a black background, which causes the flickering. I decided to find out exactly what it was that was causing the flicker, to be able to reproduce it. This took me a while. Here's a recipe: xemacs -vanilla M-x shell ls -l M-x occur RET rw RET M-x lazy-lock-mode Move mouse around in the *Occur* buffer. See gutter repeatedly flash. lazy-lock is probably doing something evil on each motion event. However, I only see this if the X client is on Linux, not on Solaris, whether using a Solaris or Linux X server? What could be the difference? X lib optimizations, perhaps. I hope you can reproduce this. If not, give me some more debugging hints. Andy> really like to know what flags in output_gutter are set causing Andy> the calls to output_display_line. In theory extents_changed Andy> should be set but not w-> gutter_extent_modiff. Unfortunately, I am so unfamiliar with the code that I don't know how to turn that paragraph into a debugging session. Perhaps I should learn the code by just reading and starting to hack? This will be slow. Andy> You did. However, I have only a dial-up connection with a max of 36kbps. Andy> And in the UK time online costs £££. I do not consider this a viable Andy> environment for debugging remotely. I keep forgetting here that some people don't have continuous mostly free internet access. My apologies.

At 05:23 PM 3/6/00 -0800, Martin Buchholz wrote: Ok, try this. I think this is the Right Fix. andy 2000-03-08 Andy Piper <andy(a)xemacs.org&gt; * gutter.c (output_gutter): add some debug. * glyphs.h (struct Lisp_Image_Instance): add display_hash. (IMAGE_INSTANCE_DISPLAY_HASH): new macro. (XIMAGE_INSTANCE_DISPLAY_HASH): ditto. * redisplay-output.c (compare_runes): use display_hash to determine if glyphs really are not the same. * glyphs.c (update_subwindow): check display_hash to see if anything really needs to be updated. If not then do nothing. Record the display_hash after updating. (image_instance_equal): compare the image_instance face also. -------------------------------------------------------------- Dr Andy Piper Senior Consultant Architect, BEA Systems Ltd

At 07:05 PM 3/21/00 -0700, Ben Wing wrote: You are right. But I'm not going to do anything about it right now... andy -------------------------------------------------------------- Dr Andy Piper Senior Consultant Architect, BEA Systems Ltd