allegro-cl archives 1998-4-9 | home index prev H thread prev K thread-next J next L |
From: CoRey Subject: Linux version of ACL, 4.3 problems Date: 1998-4-9 20:40 I have some bugs and questions to report regarding the free Linux version of acl, 4.3. BUGS: ----- First, (subseq "HellO" 0 7) will return "HELLO<GARBAGe>"!!! Where GARBAGE appears to be random noise in the computer's memory! This should never happen in LISP! You see stuff like this in C, but Lisp's run time type checking should make sure this never occurs - it should throw an error (see CLTL2) Secondly, the random-text generator program from Paul Graham's _ANSI COMMON LISP_ does no work with ACL without a change of a major function (random-next '|word|), which finds the next word to generate using the interned symbol that corresponds with that word. The function makes use of reduce with a :key argument, but ACL's reduce does not know about the :key keyword argument! Incidentally, Graham gives a version of random-next that does NOT use reduce, claiming that reduce did not have :key back in 1984. And that function is less succint than the one which uses reduce, :key. Can't you guys bring your Lisp up to date? Thirdly, I am having problems getting two large systems to make; 1. MAXIMA - based on the original MacSyma at MIT, the first symbolic mathematics program 2. STARSIM - A simulator of *LISP on the connection machine, pretty interesting. With both packages, I have all kinds of trouble getting them to make. The odd thing is, STARSIM defines a read macro that controls whether this code will be modified for ACL! MAXIMA is written for Kyoto CL, and not knowing the differneces between those two dialects, I can't seem to convert it. I plan on getting a real Symbolics account to use over the net, so that I can make use of the #+Symbolics read macros that are in those sources... I'm not fluent enough to fix the problem. I think it stems from a weird way packages are implemented; I have problems with the forms require, in-package, etc... I realize you said there are problems with the Linux version. Those are just two things I have found, and I'll report more as I gain more facility with Lisp. FEATURES --------- The emacs lisp files that come with ACL are pretty nice, but the indentation routines are not canonical; when doing an if form, the two possible conditions that come after the test are not lined up properly. I see something similar with a setf with a multiple number of setting forms. And the annoying thing is, if I indent those forms in the style of Lisp, after I hit newline, the stupid emacslisp program REINDENTS them the WRONG way!!! How do I open up more than one lisp listener? After fi:common-lisp I can do fi:open-lisp-listener but any subsequent invocations of that command only bring me to the second lisp listener. I can't seem to get more than 2 at a time... Other than that , I enjoy the product, and wish that I could afford the full version. I like the fact that I can use sockets and regexps like I am used to with unix, shell scripting. Very nice. And the threading concept is new to me, I realize there are threads in Solaris, C, etc, but don't understand how to use them yet. Is there a guide to this or a tutorial somewhere? Corey Carroll <omega.uta.edu at cjc9024> |
(
1
CoRey /
Linux version of ACL, 4.3 problems
4/9
)
(
2
Daniel Solaz/
...
4/10
)
(
3
Alex Athanasopoulos /
...
4/13
)
(
4
Daniel Solaz/
...
4/13
)
(
5
CoRey /
...
4/11
)
(
6
Arthur Flatau /
...
4/12
)
(
7
Kenneth D. Forbus/
...
4/12
)
(
8
Arthur Flatau /
...
4/10
)
(
9
Erik Naggum /
...
4/10
)
(
10
Roy Turner /
...
4/10
)
(
11
Erik Naggum /
...
4/10
)
(
12
Roy Turner /
...
4/10
)
(
13
Erik Naggum /
...
4/10
)
(
14
Erik Naggum /
...
4/9
)
|