When trying to compile cl-http for ACL I always get this error:
; Loading /pub/cl-http-67-100/clim/clim-sys/defsystem.lisp
Error: Illegal sharp character #\space [file position = 4620]
[condition type: READER-ERROR]
I can't see anything weird in the source code. Any ideas?
For CMUCL when I begin the load I get:
* (load "http:cmucl;start.lisp")
; Loading #p"/pub/cl-http-67-100/cmucl/start.lisp".
;; Loading #p"/pub/cl-http-67-100/cmucl/translations.lisp".
Argument Y is not a REAL: NIL.
Restarts:
0: [CONTINUE] Return NIL from load of "http:cmucl;start.lisp".
1: [ABORT ] Return to Top-Level.
Debug (type H for help)
(KERNEL:TWO-ARG-< 0 NIL)
Source:
; File: target:code/numbers.lisp
(TWO-ARG-</> TWO-ARG-< < FLOOR CEILING ...)
I applied the included patches but I got a few weird errors during it.
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
··············@yahoo.com ��(��) <············@nnrp1.dejanews.com> �����
�ۼ��Ͽ��4ϴ�...
>When trying to compile cl-http for ACL I always get this error:
>
>; Loading /pub/cl-http-67-100/clim/clim-sys/defsystem.lisp
>Error: Illegal sharp character #\space [file position = 4620]
> [condition type: READER-ERROR]
>
>I can't see anything weird in the source code. Any ideas?
If my memory serves me, the file has some special characters. Deleting the
special characters will fix the problem.
- jwc
"lispM" <·····@yahoo.com> writes:
> ··············@yahoo.com ��(��) <············@nnrp1.dejanews.com> �����
> �ۼ��Ͽ��4ϴ�...
> >When trying to compile cl-http for ACL I always get this error:
> >
> >; Loading /pub/cl-http-67-100/clim/clim-sys/defsystem.lisp
> >Error: Illegal sharp character #\space [file position = 4620]
> > [condition type: READER-ERROR]
> >
> >I can't see anything weird in the source code. Any ideas?
>
>
> If my memory serves me, the file has some special characters. Deleting the
> special characters will fix the problem.
By coincidence, I just downloaded this version of cl-http and tried it
in ACL 5.0 and got the same error (though my image reported a slightly
different file position--4620 is the right position; I'm still puzzling
over why my image said 4773--see below). The problem was in
cl-http-67-100\acl\start.lisp at about line 142, where there was a "# "
just like the error says. Looks like it wanted to be a ";" instead of "#".
The line in question was in save-lisp-image and said:
# hum this spawns a process to save image...
and I changed it to use ";" instead of "#".
Fixing this in the source allowed the rest of the system to compile/load
for me .. anyway, it wasn't a binary problem in my case.
your mileage may vary.
Incidentally, trying to find this problem in Emacs was made more complicated,
I think, by the fact that the file system stores CRLF as two chars, but
emacs uses only one char, so going to the start of the buffer and typing
c-u 4 7 7 3 c-f didn't take me to the right
place. I eventually figured out to subtract a couple hundred from the
file position for all those CRLF's that Emacs was treating as just returns
and it took me to the almost the right place. The file position robert cites above
is exactly what the value should have been. 4773-146=4633 gets me close. (146
because char 4773 occurs on line 146, not line 142). Even accounting for
fenceposts, I still couldn't account for the other 13 characters. If
anyone is familiar with this problem or has an alternate diagnosis, I'm
quite curious.
I don't have CMUCL so couldn't try it there, but I bet it's the same
problem. Note that the problem is inside a #+Allegro, but if that code
is touched by CMUCL, it's possible that it still blows out. The reader
still has to read failing conditionals, and still can blow out if the
syntax is invalid while doing so.
Kent M Pitman <······@world.std.com> writes:
> I don't have CMUCL so couldn't try it there, but I bet it's the same
> problem. Note that the problem is inside a #+Allegro, but if that code
> is touched by CMUCL, it's possible that it still blows out. The reader
> still has to read failing conditionals, and still can blow out if the
> syntax is invalid while doing so.
Yes, it would, but if I understand correctly, then none of the files
in the acl directory will ever be loaded when working with CMUCL, so
this problem has not cropped up here. The conditionals are in there
only for differentiating between the various Franz CL products, IMHO.
Working with CMU Common Lisp 18a+ release x86-linux 2.4.7 6 November
1998 cvs, I had no problems building and running the whole cl-http
suite in version 67.100[1].
Just be sure to read http:cmucl;read-me.text and apply the patches in
http:cmucl;cl-http-patches.text
Regs, Pierre.
Footnotes:
[1] With the exception of Lambda-Vista, which the cmucl-specific
documentation tells me is broken on implementations other than Genera
and MCL.
--
Pierre Mai <····@acm.org> http://home.pages.de/~trillian/
"One smaller motivation which, in part, stems from altruism is Microsoft-
bashing." [Microsoft memo, see http://www.opensource.org/halloween1.html]
In article <···············@world.std.com>,
Kent M Pitman <······@world.std.com> wrote:
(...)
> Incidentally, trying to find this problem in Emacs was made more complicated,
> I think, by the fact that the file system stores CRLF as two chars, but
> emacs uses only one char, so going to the start of the buffer and typing
> c-u 4 7 7 3 c-f didn't take me to the right
> place. I eventually figured out to subtract a couple hundred from the
> file position for all those CRLF's that Emacs was treating as just returns
> and it took me to the almost the right place. The file position robert cites above
> is exactly what the value should have been. 4773-146=4633 gets me close. (146
> because char 4773 occurs on line 146, not line 142). Even accounting for
> fenceposts, I still couldn't account for the other 13 characters. If
> anyone is familiar with this problem or has an alternate diagnosis, I'm
> quite curious.
Tabs (expanded/unexpanded)?
NULs? Other non-printing characters?
(I don't suppose there were any 16-bit bytes (like Unicode
characters) in the file, were there?)
(I don't have the file with me, this is just guessing.)
Vassil Nikolov <········@poboxes.com> www.poboxes.com/vnikolov
(You may want to cc your posting to me if I _have_ to see it.)
LEGEMANVALEMFVTVTVM (Ancient Roman programmers' adage.)
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
From: Howard R. Stearns
Subject: Re: Compiling cl-http for ACL or CMUCL.
Date:
Message-ID: <369BBDCD.A2B0C597@elwood.com>
A couple of points:
1. Get the patches from the CL-HTTP home page. They do fix some
problems.
2. There is a bug mailing list for CL-HTTP, and specific bug lists for
each platform. It is not clear to me, however, whether things should
be reported to the platform specific bug lists or the general one when
you run into problems. Typically, at that point, I don't know whether
I've found a platform-specific BUG or am just personally clueless. It
is also not clear if it is intended that users should subscribe to
these platform-specific lists as the only/prefered way to keep up to
date on reports, or whether they are intended more for implementers.
3. I ran into a small number of bootstrap issues trying to get CL-HTTP
to work on ACL/Linux. SOME of these have been fixed in the patches,
but I do not believe that all have been fixed...
4. I reported these to to ···············@ai.mit.edu, but never got a
response. For what it's worth, here's what I sent:
Using ACL 5 (not beta) on RedHat Linux 4.1, I encountered the following
startup problems:
acl/start.lisp:
defun save-lisp-image: # character is used as comment instead of ;
bind80: No instructions on building.
acl/-read-me-.text gives no instructions regarding building bind80.
acl/http -build: issues warnings and gives instructions which it
says are for SunOS, not linux.
acl/bind80.c: wrong cpp def.
The line to #include <rpcsvc/nis.h> is guarded by a cpp conditional:
#ifndef LINUX
Some systems define linux, not LINUX, so the line should be:
#if !((defined LINUX) || (defined linux))
acl/bind80.c: Compilation doesn't require -lsocket -lnsl on linux?
The socket and nsl libraries were not found on my linux system, but
compilation of bind80.c and 'acl/http -build' seemed to work without
it.
acl/http.script: test -x $RUNNER fails in ACL5.
The DOIT code at the end of the script checks to see if the
image is executable using 'if test -x $RUNNER ...'
However, when using ACL5, $RUNNER is something like:
/usr/local/acl5/lisp -I
/home/.../cl-http-67-100/acl/http_image.Linux.i586
This causes the test and the script to fail with:
acl/http: test: too many arguments
Changing the test to 'if test -x $HTTP/$IMAGE ...' fixes the problem.
··············@yahoo.com wrote:
>
> When trying to compile cl-http for ACL I always get this error:
>
> ; Loading /pub/cl-http-67-100/clim/clim-sys/defsystem.lisp
> Error: Illegal sharp character #\space [file position = 4620]
> [condition type: READER-ERROR]
>
> I can't see anything weird in the source code. Any ideas?
>
> For CMUCL when I begin the load I get:
> * (load "http:cmucl;start.lisp")
> ; Loading #p"/pub/cl-http-67-100/cmucl/start.lisp".
> ;; Loading #p"/pub/cl-http-67-100/cmucl/translations.lisp".
>
> Argument Y is not a REAL: NIL.
>
> Restarts:
> 0: [CONTINUE] Return NIL from load of "http:cmucl;start.lisp".
> 1: [ABORT ] Return to Top-Level.
>
> Debug (type H for help)
>
> (KERNEL:TWO-ARG-< 0 NIL)
> Source:
> ; File: target:code/numbers.lisp
> (TWO-ARG-</> TWO-ARG-< < FLOOR CEILING ...)
>
> I applied the included patches but I got a few weird errors during it.
>
> -----------== Posted via Deja News, The Discussion Network ==----------
> http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
From: Christopher R. Barry
Subject: Re: Compiling cl-http for ACL or CMUCL.
Date:
Message-ID: <874spwunrm.fsf@2xtreme.net>
"Howard R. Stearns" <······@elwood.com> writes:
> A couple of points:
>
> 1. Get the patches from the CL-HTTP home page. They do fix some
> problems.
>
> 2. There is a bug mailing list for CL-HTTP, and specific bug lists for
> each platform. It is not clear to me, however, whether things should
> be reported to the platform specific bug lists or the general one when
> you run into problems. Typically, at that point, I don't know whether
> I've found a platform-specific BUG or am just personally clueless. It
> is also not clear if it is intended that users should subscribe to
> these platform-specific lists as the only/prefered way to keep up to
> date on reports, or whether they are intended more for implementers.
>
> 3. I ran into a small number of bootstrap issues trying to get CL-HTTP
> to work on ACL/Linux. SOME of these have been fixed in the patches,
> but I do not believe that all have been fixed...
>
> 4. I reported these to to ···············@ai.mit.edu, but never got a
> response. For what it's worth, here's what I sent:
>
> Using ACL 5 (not beta) on RedHat Linux 4.1, I encountered the following
> startup problems:
>
> acl/start.lisp:
> defun save-lisp-image: # character is used as comment instead of ;
Yes, the now infamous `File Position 4620' bug.
> bind80: No instructions on building.
> acl/-read-me-.text gives no instructions regarding building bind80.
> acl/http -build: issues warnings and gives instructions which it
> says are for SunOS, not linux.
I also searched awhile for the bind80 build instructions. When I
finally looked at the source code I found them at line number 2 of
bind80.c (and they weren't correct either).
> acl/bind80.c: wrong cpp def.
> The line to #include <rpcsvc/nis.h> is guarded by a cpp conditional:
> #ifndef LINUX
> Some systems define linux, not LINUX, so the line should be:
> #if !((defined LINUX) || (defined linux))
This is the case on Debian, and was also annoying.
> acl/bind80.c: Compilation doesn't require -lsocket -lnsl on linux?
> The socket and nsl libraries were not found on my linux system, but
> compilation of bind80.c and 'acl/http -build' seemed to work without
> it.
cc bind80.c -o bind80
> acl/http.script: test -x $RUNNER fails in ACL5.
> The DOIT code at the end of the script checks to see if the
> image is executable using 'if test -x $RUNNER ...'
> However, when using ACL5, $RUNNER is something like:
> /usr/local/acl5/lisp -I
> /home/.../cl-http-67-100/acl/http_image.Linux.i586
> This causes the test and the script to fail with:
> acl/http: test: too many arguments
> Changing the test to 'if test -x $HTTP/$IMAGE ...' fixes the problem.
Wish my version of ACL could save an image. (Not that I'm really
complaining, Allegro IMHO is by far the most generous of the
commercial vendors. Compare what their Linux trial edition gives you
compared to the crippleware you get with MCL and the like or the
non-existant trial offerings of other vendors. It just gets to be a
pain to wait 40 seconds or so for CL-HTTP to load and to edit out
those questions it keeps asking you every time).
From: Robert Monfera
Subject: Re: Compiling cl-http for ACL or CMUCL.
Date:
Message-ID: <369C4E59.9C796598@fisec.com>
"Christopher R. Barry" wrote:
...
> Wish my version of ACL could save an image.
I thought you can save an image with their free Linux version. Do you
refer to
the windows version?
> (Not that I'm really
> complaining, Allegro IMHO is by far the most generous of the
> commercial vendors. Compare what their Linux trial edition gives you
> compared to the crippleware you get with MCL and the like or the
> non-existent trial offerings of other vendors.
Another major vendor is Harlequin. They give you a free LWW 4.1
Personal edition. Have you tried if it is able to load cl-http?
Their LWW 4.1 Professional version costs something I do not consider too
pricey. It even gives you royalty-free delivery. Their LWW 4.01
version also included ODBC.
Maybe you are not interested in GUI or ODBC, in which case ACL for Linux
is a great non-commercial environment that you can get for free.
HTML is still far from something I could use as a GUI, and most
customers use relational databases accessible via ODBC (these two items
and the lack of an IDE stops me from trying Eclipse as well).
Is there something I am missing here?
Okay, ACL 5 for winfows looks more sophisticated than LWW, but I do not
have enough experience with it to compare them.
> It just gets to be a
> pain to wait 40 seconds or so for CL-HTTP to load
> and to edit out
> those questions it keeps asking you every time).
Probably you can edit the source code so that you do not have to answer
every time, and you can make ACL to load cl-http on startup.
Regards
Robert
From: Christopher R. Barry
Subject: Re: Compiling cl-http for ACL or CMUCL.
Date:
Message-ID: <87pv8jfti5.fsf@2xtreme.net>
Robert Monfera <·······@fisec.com> writes:
> "Christopher R. Barry" wrote:
>
> ...
> > Wish my version of ACL could save an image.
>
> I thought you can save an image with their free Linux version. Do
> you refer to the windows version?
No, you can start the excl:build-lisp-image process but it will always
die on you.
> > (Not that I'm really
> > complaining, Allegro IMHO is by far the most generous of the
> > commercial vendors. Compare what their Linux trial edition gives you
> > compared to the crippleware you get with MCL and the like or the
> > non-existent trial offerings of other vendors.
>
> Another major vendor is Harlequin. They give you a free LWW 4.1
> Personal edition.
Not on unix. And they list about 10 different unix variants they
support but no mention of Linux among them.
> Have you tried if it is able to load cl-http?
> Their LWW 4.1 Professional version costs something I do not consider too
> pricey. It even gives you royalty-free delivery.
And how does one get a chance to evaluate it before purchasing? I
think that their lack of a trial edition for unix suggests that the're
maybe not too eager to have their product readily accessible to anyone
for comparison to competing unix offerings like Allegro.
> Maybe you are not interested in GUI or ODBC
I've got my Garnet fix for now. When Mike MacDonald's Free-CLIM .0001
comes out I'll be able to look into CLIM 2.0 then.
> in which case ACL for Linux is a great non-commercial environment
> that you can get for free.
Non-commercial? Allegro CL is a great environment period. I don't know
what you mean by non-commercial though. And you seem to imply that if
one is not interested in GUI or ODBC then ACL is a good choice. Why
wouldn't it be a good choice if one was interested in these things,
for which Allegro also has offerings?
> HTML is still far from something I could use as a GUI, and most
> customers use relational databases accessible via ODBC (these two items
> and the lack of an IDE stops me from trying Eclipse as well).
IDE? I can't think of any model of computer that supports a text-mode
monitor that Emacs doesn't run on :-)
> Is there something I am missing here? Okay, ACL 5 for winfows looks
> more sophisticated than LWW, but I do not have enough experience
> with it to compare them.
>
> > It just gets to be a
> > pain to wait 40 seconds or so for CL-HTTP to load
> > and to edit out
> > those questions it keeps asking you every time).
>
> Probably you can edit the source code so that you do not have to answer
> every time, and you can make ACL to load cl-http on startup.
I did take care of the question thing. It was just annoying that I had
to (I shouldn't have brought it up, was typing faster than brain was
functioning). As for loading CL-HTTP on startup, that would be even
worse than it is now.
Christopher
From: Erik Naggum
Subject: Re: Compiling cl-http for ACL or CMUCL.
Date:
Message-ID: <3125221966993849@naggum.no>
* ······@2xtreme.net (Christopher R. Barry)
| No, you can start the excl:build-lisp-image process but it will always
| die on you.
really? it will? perhaps you refer (well, "refer" is _overly_ generous)
to the problem of the missing "prof" file referenced by the build process
because the Linux Trial Edition believes it's an Enterprise Edition. it
has a fairly simple fix. this is the relevant line in develenv.cl:
#+allegro-cl-enterprise (require :prof) ;; not allowed in runtime images
this is what it should look like:
#+allegro-cl-enterprise (ignore-errors (require :prof)) ;; not allowed in runtime images
let me know if you still have problems.
| And how does one get a chance to evaluate it [Harlequin's LWW] before
| purchasing? I think that their lack of a trial edition for unix suggests
| that they're maybe not too eager to have their product readily accessible
| to anyone for comparison to competing unix offerings like Allegro.
I don't know about Harlequin, but Franz Inc have always had evaluation
periods for their Unix products, even before Linux existed. it is
completely unwarranted to draw conclusions from one possible answer to
the question you ask without actually giving anybody a chance to answer.
| IDE? I can't think of any model of computer that supports a text-mode
| monitor that Emacs doesn't run on :-)
VAX/VMS support was discontinued some time ago. (sorry.)
#:Erik
From: Robert Monfera
Subject: Re: Compiling cl-http for ACL or CMUCL.
Date:
Message-ID: <369CC820.41ACF815@fisec.com>
"Christopher R. Barry" wrote:
>
> > Maybe you are not interested in GUI or ODBC
> > in which case ACL for Linux is a great non-commercial environment
> > that you can get for free.
>
> Non-commercial? Allegro CL is a great environment period.
Agreed.
> I don't know
> what you mean by non-commercial though.
It is explained in the license.
> And you seem to imply that if
> one is not interested in GUI or ODBC then ACL is a good choice.
Yes, indeed it is a good choice.
> Why
> wouldn't it be a good choice if one was interested in these things,
> for which Allegro also has offerings?
Probably it is an excellent choice, but remember your original posting
was about generous vendors rather than whether or not a choice is good.
I disagreed that you made a broad statement on vendor generosity (which
is an ambiguous term here anyway).
Regards
Robert
In article <··············@2xtreme.net>,
Christopher R. Barry <······@2xtreme.net> wrote:
...
>> Their LWW 4.1 Professional version costs something I do not consider too
>> pricey. It even gives you royalty-free delivery.
>
>And how does one get a chance to evaluate it before purchasing?
On Windows? Yes. There is the personal edition.
> I
>think that their lack of a trial edition for unix suggests that the're
>maybe not too eager to have their product readily accessible to anyone
>for comparison to competing unix offerings like Allegro.
Have you asked for a "temporary" version? Just because there
is no "30 days test edition" on the anonymous ftp site doesn't mean you
cannot take it out for a "test drive". With a convincing query,
I imagine the sales folks could be amenable.
Harlequin's Lisp products will be late to the "Linux revolution".
>Non-commercial? Allegro CL is a great environment period. I don't know
>what you mean by non-commercial though.
Non commerical I think means that you cannot use the environment for
commerical activities. If you are hacking around learning lisp and/or
personal projects then it is free. If you are going to deliver
commerical products/services (e.g., an application, HTTP server ) you
need to pay.
--
Lyman S. Taylor "Twinkie Cream; food of the Gods"
(·····@cc.gatech.edu) Jarod, "The Pretender"
Christopher R. Barry wrote:
> > Their LWW 4.1 Professional version costs something I do not consider too
> > pricey. It even gives you royalty-free delivery.
>
> And how does one get a chance to evaluate it before purchasing? I
> think that their lack of a trial edition for unix suggests that they're
> maybe not too eager to have their product readily accessible to anyone
> for comparison to competing unix offerings like Allegro.
Harlequin are very happy for people to evaluate lisp before purchasing it.
Please send mail to ···············@harlequin.com to arrange this if you are
interested.
- nick
······@2xtreme.net (Christopher R. Barry) writes:
> "Howard R. Stearns" <······@elwood.com> writes:
> > acl/http.script: test -x $RUNNER fails in ACL5.
> > The DOIT code at the end of the script checks to see if the
> > image is executable using 'if test -x $RUNNER ...'
> > However, when using ACL5, $RUNNER is something like:
> > /usr/local/acl5/lisp -I
> > /home/.../cl-http-67-100/acl/http_image.Linux.i586
> > This causes the test and the script to fail with:
> > acl/http: test: too many arguments
> > Changing the test to 'if test -x $HTTP/$IMAGE ...' fixes the problem.
>
> Wish my version of ACL could save an image. (Not that I'm really
I sent several proposals to ···············@ai.mit.edu about incorrect
behavior of cl-http building script (acl/http) and startup file
(acl/start.lisp) with ACL 5/Linux. Some of them are implemented in
final 67.100 version but some didn't get there. So it still doesn't
work. You can take my version of acl/http script from
http://www2.cybercities.com/s/sparrow/http
It allows to compile, build and run cl-http image with ACL 5/Linux.
Regards,
Alexander
--
People who like this sort of thing will find this is the sort of thing
they like
Abraham Lincoln