NICK 17-Jun-85 21:37:35 Tape Software - lack thereof You really should have some interface to the tape drive better than just fs:copy-directory. Being a UNIX person I would naturally suggest that you scrap any CADR-like tape system and just implement tar in lisp. I won't even discuss how bad fs:restore-magtape is. The transform argument is just too damn low-level. It took me a while to write a half-way decent interface - most users won't screw with it at all. I'm certainly not suggesting that you copy SLIME and their five tape formats. But you need something better than this. NICK 17-Jun-85 21:39:34 Too many 1's in etcttys If you turn on all the ports in /etc/ttys, it hangs unix (and the lambda too). Now there are some there that are for stuff I don't know about, but a more graceful failure mode would help. NICK 17-Jun-85 21:46:15 Indices GET A PERMUTED INDEX! And get rid of the various 'concept' and 'function' etc. indices in the LMM. Put everything into one permuted index per volume, and one volume that is just permuted index. And for some reason the page numbers in the indices are frequently wrong ... NICK 17-Jun-85 21:47:52 missing documentation There are no unix 'man' page descriptions (or any documentation at all, as far as I can see) for telnet, cftp, etc. The on-line stuff is good but it should be duplicated elsewhere. NICK 17-Jun-85 21:50:39 unix Why are you using AT+T unix anyway? 4.2 is the acknowledged standard; System V 'now has all the Berkeley enhancements'. V7 is the pits. At the SUN user group convention in Boston (Nov 84), Bill Joy was on a panel with a fellow from DARPA and one from AT&T. Someone asked the DARPA guy if, since DARPA had funded 4.2, DARPA would have trouble funding any project that proposed using AT&T unix. The DARPA fellow said "Well, I'm not sure I know how to answer your question, since the situation has never arisen ..." 4.2 is the best. NICK 17-Jun-85 21:55:02 apropos (apropos 'pathname-def) blows up, yielding an error message. Apropos should catch almost any conceivable error. NICK 17-Jun-85 21:56:11 dired s command The s command in dired should say 'empty directory' if the selected directory really has no subfiles, instead of ignoring you. The second time it is typed it should just remove the subfiles without asking. Or at least change the prompt to 'remove subfiles from screen' so you don't have to explain how it doesn't *really* delete files ... NICK 17-Jun-85 21:59:15 directory creation fs:create-directory should be called a lot more often. Using a pathname should quietly and automagically create the appropriate subdirectories. See tar. NICK 17-Jun-85 22:00:35 dired Why can't I delete empty top-level subdirectories in dired? NICK 17-Jun-85 22:27:01 file-system Something needs to be done about directory merging. I end up typing a full dirspec far too often. Maybe something equivalent to the unix '..' meaining 'up' would help. at the very least I should be able to merge: lama:nic.foo.bar;test.lisp with .blort.baz; to yield lama:nic.blort.baz;test.lisp Obviously there is some ambiguity here. Did I mean nic.foo.bar.blort.baz instead? But this is no excuse for making me type full dirspecs all the time. Why can't a lispm have a concept of 'current directory'? Or are my unix colors showing through? The lispm is supposed to have the el-primo user interface. How about filename completion with a menu of possible completions, and not just in Zmacs, either? NICK 17-Jun-85 22:33:20 methods and c-sh-a Lisp Listeners should have the ability to do ctrl-sh-a on flavor methods, just like Zmacs can. It can't be *that* much overhead. Ztop is NOT a solution. NICK 17-Jun-85 22:34:20 lisp listeners Lisp Listeners should scroll up and down like a normal display terminal. The blitting isn't that slow. Keep characters scrolled off the screen in 8-bit semi-ascii representation, not as bit-maps, to save space. Allow the window to be scrolled forward and backward. Allow characters to be 'yanked' off the screen with the mouse, aka SUN. (hmm... maybe you should scroll windows left and right too ...) NICK 17-Jun-85 22:37:24 window system It is a fallacy that de-exposed windows cannot update themselves. I should be able to have two overlapping windows, both printing out chars or graphics, with the keybd stream connected to only one, and the appropriate stuff happening. See how SUN does it. The capability is frequently useful. Having 'System-x' keys does not compensate. NICK 17-Jun-85 22:42:24 mice For 10K from SUN I can buy a system that does full mouse-ahead. X-Y position should be read when the button is clicked, not when the processor gets around to noticing this. If you need special hardware, so what? NICK 17-Jun-85 22:44:35 scroll bars I know you must hate xerox scroll bars, but most people find them a hell of a lot easier to use ... NICK 17-Jun-85 22:45:25 ztop ztop mode takes a ctrl-meta-y, but not a ctrl-c. Lisp Listeners take both. Slime uses c-m-y, ctrl-c is what's documented. Which shall it be? NICK 17-Jun-85 22:46:25 asembly lang There is a fairly major barf in the LMM, page 756. Apparently the compiler got smarter and bummed out the MOVE instruction talked about in para 3. This area is full of lots of other errors ... Maybe no one reads assembly any more ... NICK 17-Jun-85 22:48:20 viewing a file The function that views files from dired should accept a at the end of the file as meaning the same as a . I.e. exit, go back to dired. SLIME users get all confused ... The extra functionality, especially the slow-scroll, is appreciated, and this is just one small little prob ... LIZ 21-Jun-85 19:27:08 in-package In-package has a local variable called pkg that is set to the named package if it is found. If the package doesn't exist, it is *not* set to the named package as it should be since the very next thing it does is a pkg-goto on pkg. Look at the code for in-package for more info! LIZ 21-Jun-85 20:38:41 lisp package The lisp package does not :use global as it should. Therefore, any (and all) packages that are supposed to be in common-lisp should do :use (lisp #+LMI global). NICK 22-Jun-85 23:20:29 Of mice and men Tv:mouse-speed-hack needs to take into account the possibility that the mouse process was swapped out between blips. Very frequently, when my machine is QFILE or TELNET serving, the mouse becomes next to unusable, because at arbitrary points the mouse process stops listening, and then wakes up when the mouse has moved miles away. Because of the 'warp' factor, the mouse then moves even FURTHER away, in sometimes unpredictable directions. Saying the machine should be single-user is not an acceptable excuse. It IS supposed to multiprocess. This situation is further aggravated by the apparent fact that several system functions deliberately warp the mouse away from where it should be. In general, I like tv:mouse-speed-hack. But before it becomes really usable, this needs to be fixed. NICK 22-Jun-85 23:45:38 disk status I've been getting an awful lot of "unexpected disk space status in CHANGE-MAP-DISK-SPACE" errors. NICK 30-Jun-85 23:52:29 r option in dired renaming a file - if you are displaying subfiles with the 's' command, and you rename a file that is underneath another (i.e. was shown by the 's' command), it gets confused and thinks it can't display the new file name (if it should be able to). It says something like 'renamed to , in a directory not in this display - but it *is* in the display. M-x revert buffer will display it. Dired needs to be smarter about files displayed with the 's' command. NICK 1-Jul-85 20:57:00 bitblt Bitblt is smart about running out of space in the to-array (i.e. if it hits the right or bottom edge. It is not so smart for the top and left edge. (i.e. negative args). Yes, you can write a handler for the blt error that re-indexes into from-array, but the bitblt itself should be smarter. ISRAEL 7-Aug-85 14:57:01 zmacs prob When you have 2 zmacs windows (i.e. 2 windows) open, using the scroll bar to move point to the top of the file in the top window, (while the cursor is in the bottom window) causes an error (using the middle mouse button percentage-wise). It works as long as you don't try to move to the top ... It also works in the bottom window, when the cursor is in the top. CLAIRE 27-Aug-85 16:32:40 The lambda and the UNIX box crashed, clock stoped, about the time she typed (si:%halt) The lambda and the UNIX box both crashed so bad that the lambda clock stoped updating and UNIX ignored all input. Claire observed that it seemed to happen because she used (si:%halt). nick 28-Aug-85 16:54:54 SETUID-IN-UNIX-DOESNT-WORK Tried it several times. It just gets ignored. In shell scripts at least, haven't tried it for a program .... nick 28-Aug-85 17:09:37 SETUID-IN-UNIX-DOESNT-WORK-2 Actually, turns out it *does* work for programs ... did some security-conscious person (contradiction in terms!) decide to flush setuid shell scripts ??? NICK 30-Sep-85 03:31:59 mouse-select process If you click left on the border of a menu, the mouse select process tries to select that menu and errors out, saying 'is this an error?'. Slimes don't do this. NICK 30-Sep-85 03:34:05 who-line updates The function that updates the who-line uses eq (actually neq) to check whether the new value is the same as the old. This probably should use equal or maybe even string-equal. JEFF 7-Nov-85 11:18:57 SUPER-BCREATE-DIRECTORY restart doesn't work When I tryu to save a file that is in a directory that doesn't exist I get an error. One of the resume possibilities is to create the directory and try again. Unfortunately this doesn't actually create the directory.