summaryrefslogtreecommitdiff
path: root/etc
diff options
context:
space:
mode:
authorGlenn Morris <rgm@gnu.org>2020-12-18 08:55:06 -0800
committerGlenn Morris <rgm@gnu.org>2020-12-18 08:55:06 -0800
commit8d2d8d7f53055d93ed2d8a435533c9a6a6303725 (patch)
treede620148368bc681a6d21912b4d39543a96e7e1a /etc
parentfdaaf886b71fc41d0d6d717af55e4927ed4cd2c1 (diff)
parent48b9c47805fc304441017f6ee4c114212cdb0496 (diff)
downloademacs-8d2d8d7f53055d93ed2d8a435533c9a6a6303725.tar.gz
Merge from origin/emacs-27
48b9c47805 Minor fixes in authors.el and in tarball-making instructions d7a4ceaa1e ; Add a new item to TODO 64fe805b19 Improve documentation of 'query-replace' 7cacf5da47 Update to Org 9.4.3 # Conflicts: # admin/authors.el
Diffstat (limited to 'etc')
-rw-r--r--etc/TODO36
1 files changed, 36 insertions, 0 deletions
diff --git a/etc/TODO b/etc/TODO
index 8e93e7fb10a..08f851076cf 100644
--- a/etc/TODO
+++ b/etc/TODO
@@ -536,6 +536,42 @@ This should go with point, so that motion commands can also move
through tall images. This value would be to point as window-vscroll
is to window-start.
+** Make redisplay smarter about which parts to redraw
+Currently, redisplay has only 2 levels of redrawing: either it
+redisplays only the selected window on the selected frame, or it
+redisplays all the windows on all the frames. This doesn't scale well
+when the number of visible frames is large.
+
+Currently, two variables are used to make the decision what to
+redisplay: update_mode_lines and windows_or_buffers_changed. These
+are set by various functions called from Lisp, and if redisplay finds
+one of them to be non-zero, it considers all the windows on all the
+frames for redisplay.
+
+The idea is to make the decision which parts need to be redrawn more
+fine-grained. Instead of simple boolean variables, we could have a
+bitmapped variable which records the kinds of changes done by Lisp
+since the previous redisplay cycle. Then the decision what exactly
+needs to be redrawn could be made based on the bits that are set.
+
+For example, one reason to consider all frames is that some scrolling
+command sets the update_mode_lines variable non-zero. This is done
+because the frame title, which doesn't belong to any window, needs to
+be reconsidered when the selected window is scrolled. But considering
+the frame title doesn't have to redisplay all the other windows on the
+frame, doesn't need to recompute the menu items and the tool-bar
+buttons, and doesn't need to consider frames other than the selected
+one. Being selective about what parts of the Emacs display need to be
+reconsidered and redrawn given the changes since the last redisplay
+will go along way towards making redisplay more scalable.
+
+One way of making this change is to go through all the places that set
+update_mode_lines and windows_or_buffers_changed, figure out which
+portions of the Emacs display could be affected by each change, and
+then implement the bitmap which will record each of these affected
+display portions. The logic in redisplay_internal will then need to
+be restructured so as to support this fine-grained redisplay.
+
** Address internationalization of symbols names
Essentially as if they were documentation, e.g. in command names and
Custom.