macrakis@UUNET.UU.NET (Stavros Macrakis) (10/19/88)
The following messages are all about Emacs version 18.50. They are
all suggestions rather than bugs.
Occur-mode-goto-occurence should save old dot in dot buffer.
My terminal supports 132 lines of 80 characters. It displays text at
about 1200 characters/sec (10kbaud). A full screen display thus takes
a long time, so it is best to avoid it. It does not support line
INS/DEL. All this taken together means:
1. When temporary windows (help, completion, ...) are put up, they
should not replace existing large windows, because that means a
lot of redisplay later.
2. When a window is suppressed (c-X 0), the other windows should not
all be resized--which would redisplay all their contents.
Instead, its space on the screen should be merged with an
existing window, normally the next one down, unless it's the
last, in which case it should be merged with the previous.
3. Temporary windows should probably appear in a fixed place on the
screen (perhaps the last 20 lines), so that you don't have to
think twice about where to look.
The other complication on my terminal is that the 132 lines appear in
two columns. As a general rule, it is desirable that windows stay
within one of the two columns (except when you have only one window).
This has implications both for window splitting and for window
suppression (in fact, it is another reason for wanting a killed
window's lines to be tacked onto an adjacent window -- preferably in
the same column -- rather than resizing all the other windows). Is
anyone else working on these problems, or should I look at them
myself?
What is case-fold-search supposed to mean for meta characters?
There doesn't seem to be any way to permanently set the minibuffer
window height to > 1. On very large screens, for users who do a lot
in the minibuffer, this might be handy.
For sites such as ours, where some machines share file systems (NFS
etc.) but others are only connected via FTP, it would be handy if the
directory structure kept .elc files that are autoloaded at runtime (as
opposed to init) in a separate directory. e.g. /lisp, /src, /load
There should be a looking-at for strings (not regexps).
----------------
For the record, here are some bugs I noticed in 18.41 which appear to
have been corrected in 18.50:
BUG: In regexp searches, ranges including meta characters don't work.
In Regexp I-Search, I tried both M-s [ C-Q 2 0 0 - C-Q 3 0 0 ] and
also M-s [ \ 2 0 0 - \ 3 0 0 ] and neither one works. On the other
hand, M-s C-Q 2 0 0 does work for the single character \200.
(case-fold-search = t)
BUG: The regexp range [@-A] matches all the following characters: @ A
[ \ ] ^ _ ` a \320 \330 \370. (case-fold-search = t)
BUG: The regexp .[@] matches all xy where y is one of @ \310 \320 \350
\370. (This is only partially reproducible. Is some table not being
initialised properly? Is someone indexing outside its bounds?)
However, [@] only matches @. Therefore this is a bug, not a feature
of case-fold-search. (case-fold-search = t)
BUG: (condition-case x (error "foo")) goes into an infinite loop (not
C-g'able).