"David A. Panariti" <davep.xemacs(a)meduseld.net> writes:
XEmacs 21.5 (beta24) "dandelion" [Lucid] (i686-pc-linux,
Mule) of Thu Dec 29
2005 on vilya
Fixes a bug in completion buffers that prevented me from using the keyboard
to move between completions. Almost everything ended up putting point at the
end of the last completion (and at point-max) It seems that something rather
fundamental has changed. Completion mode is derived from list-mode.
list-mode uses properties to track/locate list items (it seems, I don't fully
understand the code.) The next and previous functions worked by moving point
until the properties changed. The properties put on the items were not text
properties but the functions used to move over properties were ones that only
worked with text properties. I changed the functions called to ones that
work with non-text properties.
Documentation for these functions is the opposite of what I'd expect:
--- c:\DOCUME~1\AICHNE~1\LOCALS~1\Temp\ediff4xA5Oc 2006-03-19 12:46:25.468750000 +0100
+++ c:\DOCUME~1\AICHNE~1\LOCALS~1\Temp\ediff4xAGZi 2006-03-19 12:46:25.484375000 +0100
＠＠ -1,4 +1,4 ＠＠
-`previous-single-property-change' (buffer: *Article*, mode: Article)
+`previous-single-char-property-change' (buffer: *Article*, mode: Article)
＠＠ -21,5 +21,5 ＠＠
the value of PROP. (Note that this situation will not happen if you always
use the text-property primitives.)
- This function looks only at extents created using the text-property primitives.
- To look at all extents, use `next-single-char-property-change'.
+ This function looks at all extents. To look at only extents created using the
+ text-property primitives, use `next-single-char-property-change'.
Who can shed some light on this?
As a point of information for me: this is dumped lisp code for a beta
version. Should this be posted here or to xemacs-beta?
xemacs-patches is the right place for all your patches you think are
ready for review and commit as-is.