09/21/87  emacs 
Known errors in the current release of emacs.
#	Associated TR's
Description

386  phx20768
Emacs gets an error if the height of the user_i/o window is reduced
during a video system Emacs session (e.g.  doing ^Z^Z and shrinking the
window for a status line).

385  phx20663
If the user lacks w access to an output file, and answers "no" to the
"Do you want to write to the protected file <path>?" query, a confusing
error message, containing only [wd]>nil, is displayed.

384  phx20641
If the current line fills the entire window, ^V (next-screen) appears
to do nothing.  This is because it tries to maintain a line of overlap
between the old and new screen, and that overlapping line happens to be
the entire old screen.

383  forum.2730
Redisplay does not do the right thing if multiple buffers are on
display in multiple windows, and the values of tab-equivalent differ
among the buffers.

382  phx20344
If accept-messages-path was used to accept messages on a mailbox other
than the user's default, the Emacs message handler is not removed from
that mailbox when quitting from Emacs.  If a message is then received
in the mailbox a null pointer fault will result.

381  phx20346
reverse-regexp-search for "^string" goes to the beginning of the
previous line containing "string" anywhere, rather than a line
beginning with "string".

380  phx20282
"esc-X help" refers to >documentation>subsystem>emacs_dir instead of
>documentation>subsystem>emacs.

377  phx19966
editor-main-init setq's default-fill-column and default-comment-column,
thus overriding the defaults that are specified in e_option_defaults_.

375  phx19737
The sequence:
          emacs -ns
          ^X-3 esc-? ^E esc-R
 results in an Emacs internal error followed by a fault referencing a
lisp temp segment, which recurs if you try to pi back in.

374  phx19730
The previous-command variable is not maintained properly.  This can
cause problems for commands that change their action when invoked
consecutively (e.g.  rubout-word and complete-command).  Commands which
throw to top-level rather than returning normally (such as the
minibuffer termination key), and commands which are on
*transparent-commands* leave previous-command unchanged, so the
affected commands think that they have been invoked consecutively when
they have not.

373  
complete-command (esc-SPACE) will get into a loop returning a sequence
of the same completions if the completion-list contains a duplicated
element.

372  
Fundamental/.ext-commands contains "alm-mode" twice, which causes error
373 to occur.

369  
"destroy-buffer-contents" should release the buffer's temp segments.
This is an easily-solved, special case of the general problem described
in entry 368.

368  
Emacs never reclaims the space in its buffer temp segments that becomes
unreferenced after a line gets copied into a lisp string, until the
buffer is destroyed.  Perhaps a GC daemon could be created that would
zero the unreferenced lines in the temp segs (in the hopes that an
entire page might be zeroed).  Or perhaps the routines that replace a
filecons with a string or delete lines could keep track of the amount
of wasted space, and trigger a compaction of the temp seg when this
passes a specified threshhold (the GC daemon cannot compact because it
might be invoked after a pseudo-pointer has been extracted from a
filecons).

367  phx19619
The internal documentation for the "replace" command says that it will
not prompt for the second string if the first isn't found.  This is not
true.

366  phx19617
multi-word-search (^XW) should start its search at the end of the
current word if it is in the middle of a word.  Otherwise it might
treat the remainder of the word as a whole word.  For instance, if
point is between the e and w in "foreword", "^XW word" will stop at the
end of "foreword".

361  phx19374
accept-messages-path references an undefined function
console-messages-interrupt-handler (it should be
console-message-interrupt-handler).

360  phx19218
Switching windows (i.e.  ^XO) before the first redisplay can cause a
buffer named "nil" to be created.  It is difficult to kill this buffer.

359  phx19170
e_cline_'s any_other handler doesn't ignore stringsize like
default_error_handler_ does.

358  phx19119
Recursive command-quit's cause problems.  At top-level it will cause a
"throw can't find catch.  pgazonga" infinite loop.  In a minibuffer it
will cause the minibuffer to be aborted and quits to be ignored.  To
reproduce:
  esc-esc unwind-protect (command-quit) (command-quit)

355  forum.2414
e_mail_pl1_$emacs_read_message returns an error code if it can't send
the cknowledgement.  It should ignore this, as read_mail does.

354  forum.2414
If RMAIL gets an error reading or formatting a message "nil" is
displayed instead of the line count and message.

353  phx18981
The electric-pl1-mode function overrides bindins of :  and ; that are
set in the pl1-mode-hook.

348  phx18717
^X^E should make sure that the file_output buffer is modifiable.

345  forum.2233
The list *transparent-commands* should include control-prefix when
vt1xx_keys_ is loaded.

339  phx18154
LDEBUG esc-R (ldebug-reset-break) does not check that the break
s-expression is still in the source buffer before deleting it.  If the
user manually deletes that s-expression then it will delete the
following sexp.

333  
If you specify a pathname as the terminal type (with -query or -ttp)
you must explicitly give the .ctl suffix.

330  phx17744
^X_ (underline-region) underlines newlines.

329  phx17727
Lisp mode incorrectly indents the third line of this construct:
 (a
   '(,b
     c))

This seems to be due to the asymmetry of forward-sexp and backward-sexp
when there are prefix read-macros around.  Some fixes are in [aref
1087] in Emacs meeting.

328  
The short name for -shared_static (undocumented) should probably be
-ss, which is what the run_command_task command uses.

325  phx17524
Buffer editor doesn't recognize buffer names (when implementing G
request, for instance) that have a ">" past the 20th character.

324  phx17333
If Emacs gets an RQO while writing out a file it doesn't always reset
the terminal properly, because it doesn't run the DCTL-epilogue.  In
the case of a VT100, for instance, the scroll region is not reset.
tty_ modes, however, do get reset properly.

322  phx17317
When RMAIL re-uses a reply buffer it doesn't make sure that the "Ack"
minor mode is reinitialized to match the acknowledgement flag.

320  
There are paths through e_tasking_ which will cause a
return_conversion_error.  See the "signal program_interrupt" statement;
if it returns then it falls through to an incompatible entrypoint.

319  phx17174
Type ^X` test.test CR CR
     ^X` noproj CR CR
 pretty soon you will get "lisp:  car or cdr of a number" loop.

313  phx16932
Emacs doesn't recognize sentences if the full stop is followed by ) or
".

311  phx16888
The "A new message has arrived" message has a 50% chance of being
overwritten by the "Do you wish to quit?" query.  Perhaps making the
message a local display would work better.

310  phx16888
If a minibufffer-print is followed by a minibuffer-response then the
question will overwrite the previous output.

309  phx16877
Emacs self-doc for delete_char says that it moves the rest of the line
"one to the right", which should be "one to the left."

308  phx16872
If the response to the terminal type query is all numbers then Emacs
just says "Do not understand.  Try again" instead of going to the "list
of types" query.

305  phx16606
Macro edit parser does not understand negative numeric arguments.  The
function macedit-digitp should recognize initial hyphens.

302  
It doesn't work to capitalize (esc-C) an underlined word.

294  mail.2.11
edit-dir doesn't understand directories that have no files.

287  phx16787
If you ^G out of the "Do you wish to quit rmail?" question after
deleting the last message the message is marked for deletion but stays
on display.

286  phx16785
Repeat-last-message (^X~) uses the last recipient if I have sent a
message since receiving one.

284  mail.4.83
The Emacs lisp error handler doesn't interact with errset properly.
"esc-esc errset (error 'foo 'bar 'fail-act) nil" should just return
nil, but it prints the error message.

273  mail.4.46
esc-^B (backward-sexp) goes into an infinite loop if there is a
semicolon ahead of point on the line.

270  mail.4.30
runoff-fill-region only works if the point is at the end of the region,
not at the beginning.

263  
If you mark a buffer for deletion in buffer-edit, and kill it with ^XK
before exiting buffer-edit then you get serious errors.

258  phx16454
It is difficult to kill buffers whose names end in whitespace, as ^XK
and ^XB rtrim the buffer name.

240  phx15723
The dg132b CTL doesn't provide enough padding for insert/delete lines.
At higher speeds it may be better to just redraw the screen.

239  
When Emacs signals quit the resetwrite can cause the initial string or
epilogue string.  It should do its own resetwrite when the user hits
BREAK, and then it should signal quit with an info structure that
indicates that no automatic resetwrite should occur.

233  phx15433
global-print (^XS) leaves out some lines if the search string contains
a newline.  This is because its main loop goes to the beginning of the
next line before each search; it should only do this in the one-line
case.

228  phx15349 phx09740 phx13412
The initial redisplay takes a fault if the buffer is looking at a large
segment.

227  phx15350
If Emacs takes a fault, it leaves the user in rawi mode.

226  phx15228 phx20036
You can't send mail from xmail if you have a tasking Emacs, and thenyou
cannot get back into the Emacs.

224  phx15059 phx16507
Scroll-current-window (^Z^V) allows the cursor to be on the bottom line
of the window when the line should wrap.  This gets redisplay confused.
Scenario:
 Go into two-window mode, with the top window having seven lines.
 Read in a file with several (>8) wrapping lines in a row.
 esc 6 esc G esc - ^Z^V <= gets into the above state
 esc - ^Z^V <= shows how confused it is
 ^X O <= further confusion f redisplay

217  phx14593
In overwrite-mode, backslash still inserts.  So does ^Q, but this is by
design.

216  phx14519 phx16784
Commands implemented using word-operator (esc-C, esc-L, and esc-U) do
not work properly if point on the whitespace preceding the first word
of the buffer.

215  phx14519 phx17565
Commands implemented using the word-operator macro (esc-C, esc-L, and
esc-U) do not work properly if there is no word in the buffer (or
perhaps if the buffer is empty).  A fix is given in the TR.

213  phx16399
Tasking Emacs should die more gracefully if you turn the video system
on or off under it.  This also applies if you turn on/off the video
system under a suspended normal Emacs.

211  phx14400
Real underlining display loses a letter if a capital letter is
underlined twice.

206  phx13052 phx15086 phx15464
Sometimes gets error_table_$io_no_permission on reconnection.  This is
probably due to getting the same devx.

205  phx12602
The window editor sometimes overwrites another buffer with the
window-editor information.  This happens if you delete the window in
which the window editor is running.

202  
In Multics-mode gets rid of the error_output switch.  This is due to
the loss of syn_$syn_attach_.

198  phx13758
^V and esc-V are confusing when redisplay turned off.  They can also
get confused if they are typed ahead.

194  phx13572
Emacs guesses that some terminal types are printing-terminals without
checking for a CTL first.

191  phx13136 phx13722
Minibuffer output isn't always folded when it runs over the
screenlinelen.

188  phx13062 phx14762
Some searching routines do not set gratuitous marks properly, causing
^W to wipe the wrong text.

185  phx12887
Dired buffers that are modified should stop ^X^C from exiting w/o query.

182  phx12653
^XO doesn't do anything with redisplay off.

180  phx12556
check-entry-name in e_multics_file_ should call check_entryname_, not
check_star_name_$path.

167  phx12166
A minibuffer message occurring during a local display sometimes aborts
the local display.

162  phx11972
indent-rigidly should redo tabification in line.

158  phx02711
Speedtype does not accept >4 character abbreviations, and setab command
does not diagnose either.

157  phx03081
pop-up windows/message mode problems.

156  phx11749
Local displays generated while in minibuffer don't get flushed.

154  phx11779
ESC-U/ESC-C/ESC-L should know how to correct _ canonicalization within
word when case-changing.

153  phx11755 phx11964
Multiple windows/set-screen-size blows away windows on ambassador
because of deficiency in DCTL code.

144  phx11398
LDEBUG breakpoints leave you in buffer where they occur, without LDEBUG
bindings, and prints message into current buffer.

132  phx10148
Multics Mode with Process buffer showing in two windows (with differnt
current points) confuses window manager about current window.

130  phx09878
Word definition seems a little screwed up in ^Z_ and ESC_ and ESC-T.

128  phx09256 phx19080
Single-character procedure names at begin of new pl1 file in electric
PL/I mode cause ":" electric to beep and not do its job.

117  phx10073
Emacs should diagnose illegal keystring ^Z-- (two minuses).

109  phx09931
esc-/ ^$ CR on blank line doesn't find next blank line.

104  phx09863
Misspellings and bad grammar in extensions.info.

92  phx09539
In fortran mode, with free-format fortran, positioning to source
lines should use fortran's line numbers, not Emacs'.

89  phx09343
Defcom for untabify is binding tab-equivalent to a string.

87  phx09365
Lines longer than fill column make CR in message buffer fail.

86  phx03655 phx10392
Emacs ^X^M qedx <CR>, BREAK, pi, q, pi sometimes fails in lisp_oprs_.
(Gets segfault or mme2 error).  Might be lisp bug.

84  phx09201
ESC-Q leaves a blank line with fill prefix after a paragraph just filled.

83  phx09092 phx09256 phx12741 phx15394
Emacs doesn't handle colons well for label arrays in electric
PL/1 mode (specifically try "action (4):").  Semicolons also not
handled well after declarations.

There is a fix for the label array bug in Emacs forum [1604].

81  phx08766
Emacs does not observe MAP mode.

78  phx08592
Emacs needs to know which conditions should be ignored, and which passed
through, and needs to pass conditions like "endpage" through.

77  phx08295 phx08925 phx12693
^C doesn't do right thing after ^XE.

75  phx06948
Multiple ESC-^Q cause strange interaction with redisplay; sometimes
paragraph creeps up and off screen, sometimes causes unnecessary
redisplay.

74  phx08227
e_pl1_$get_mcs_tty_info appears to not know that diablo1620's does
have tabs (and doesn't direct Emacs to use tabs in printing-tty output).

69  phx08002
When in electric-pl1-mode, ";" inside string should cause "Unbalanced
PL/1 quotes" but not newline/special action, just insert ";".

68  phx07939
Local keybindings need to be "pushed"; when do electric-pl1-mode,
fill mode, fill off, SEMI/TAB get rebound to self-insert instead
of pl1-mode-special-whatever.

63  phx07303
LRU minibuffer-print line-selection algorithm needs to be better.

61  phx07287
Emacs should do resetread on quit.

59  phx06969
On printing terminal (TI745) "Writing" got overwritten by
pathname -- possible padding bug in TI745 ctl.

52  phx06832
Marks not always moved correctly on ^W; see TR.

47  phx06696
Setting minibuffer size resets screensize.

45  phx06634
When a major-mode change happens, the OLD major mode should be
cleaned up before the new one is instated.  In particular, trying to
use an old fortran-mode buffer in most other major modes
will still find CR bound to fortran's return-and-indent.

44  phx06599
ESC-W (copy-region) doesn't respect ESC-^W (doesn't append to
last deletion.)

43  phx06588
pl1-comment-current-line should obey comment-column, NOT maintain its own
comment column (pl1-comment-column).

36  phx06301
Rmail bugs: (1) Should do a go-to-or-create-buffer "Incoming Mail", not
go-to-buffer. (2) rmail.info needs to be installed. (3) The feature
whereby ^XR in an emacs with a non-quitted rmail returns to reading
the mail you were reading MUST BE Documented.

29  phx05495 phx05675 phx06167
Speedtype mode and speedtype/fill interaction have several bugs.
Bill York's new speedtype should be installed.

18  phx03094 phx13763
On tek4025, Emacs should first ask user to set command character.
Would be nice to have a real cursor, also.

16  phx05086 phx06698 phx08108 phx11313 phx11397 phx13319
Emacs gives useless error message on RQO - should try to clean
 up pdir automatically if that's where the error is.

15  phx05073
accept-messages doesn't work in recursive Emacs'.  This is caused by
the console-message package not re-instating old accept_messages state;
this will be fixed when the new message routines roll around.

13  phx04933
Bugs in ^XR after reading another mailbox, then trying to read own.
See TR.

8  phx06674
^X:'ing to a person when that recipient's message buffer has been
deleted blows out badly. (Two error messages, including one from
go-to-line-point).

6  phx04679 phx09752 phx14625
Windows displaying the same buffer sometimes don't keep their own
curpoint; moving in one, then ^XO or ^X4 to the other sometimes
causes the second to redisplay to the point of the first.


                                          -----------------------------------------------------------


Historical Background

This edition of the Multics software materials and documentation is provided and donated
to Massachusetts Institute of Technology by Group BULL including BULL HN Information Systems Inc. 
as a contribution to computer science knowledge.  
This donation is made also to give evidence of the common contributions of Massachusetts Institute of Technology,
Bell Laboratories, General Electric, Honeywell Information Systems Inc., Honeywell BULL Inc., Groupe BULL
and BULL HN Information Systems Inc. to the development of this operating system. 
Multics development was initiated by Massachusetts Institute of Technology Project MAC (1963-1970),
renamed the MIT Laboratory for Computer Science and Artificial Intelligence in the mid 1970s, under the leadership
of Professor Fernando Jose Corbato. Users consider that Multics provided the best software architecture 
for managing computer hardware properly and for executing programs. Many subsequent operating systems 
incorporated Multics principles.
Multics was distributed in 1975 to 2000 by Group Bull in Europe , and in the U.S. by Bull HN Information Systems Inc., 
as successor in interest by change in name only to Honeywell Bull Inc. and Honeywell Information Systems Inc. .

                                          -----------------------------------------------------------

Permission to use, copy, modify, and distribute these programs and their documentation for any purpose and without
fee is hereby granted,provided that the below copyright notice and historical background appear in all copies
and that both the copyright notice and historical background and this permission notice appear in supporting
documentation, and that the names of MIT, HIS, BULL or BULL HN not be used in advertising or publicity pertaining
to distribution of the programs without specific prior written permission.
    Copyright 1972 by Massachusetts Institute of Technology and Honeywell Information Systems Inc.
    Copyright 2006 by BULL HN Information Systems Inc.
    Copyright 2006 by Bull SAS
    All Rights Reserved