From: Stig E S. <st...@ii...> - 2002-02-21 16:40:26
|
hei, I have some troubles getting REINITIALIZE-INSTANCE to work in CLISP. The following test-case: (in-package :cl-user) (defclass component () ((name :type string :accessor component-name :initarg :name) (version :accessor component-version :initarg :version) (in-order-to :initform nil :initarg :in-order-to) (inline-methods :accessor component-inline-methods :initform nil) (relative-pathname :initarg :pathname))) (defun upd-inst (obj &rest args) (apply #'reinitialize-instance obj args)) ;;(trace reinitialize-instance) (let ((foo (make-instance 'component :name "gib"))) (warn "Going 1") (upd-inst foo :name "gah1") (warn "Going 2") (upd-inst foo :name "gah2") (warn "Going 3") (upd-inst foo :name "gah3") ) Gives me: $ clisp -q -ansi reinit.lisp WARNING: Going 1 WARNING: Going 2 WARNING: Going 3 *** - handle_fault error2 ! address = 0x4556454F not in [0x202A6000,0x2036220C) ! SIGSEGV cannot be cured. Fault address = 0x4556454F. Segmentation fault (core dumped) This is quite problematic, and it's not a good thing that clisp crashes and dumps core. Anyone? -- ------------------------------------------------------------------ Stig Erik Sandoe st...@ii... http://www.ii.uib.no/~stig/ |
From: Sam S. <sd...@gn...> - 2002-02-23 18:02:19
|
> * In message <200...@ap...> > * On the subject of "[clisp-list] REINITIALIZE-INSTANCE problems" > * Sent on Thu, 21 Feb 2002 17:40:12 +0100 > * Honorable Stig E Sandoe <st...@ii...> writes: > > *** - handle_fault error2 ! address = 0x4556454F not in > [0x202A6000,0x2036220C) ! > SIGSEGV cannot be cured. Fault address = 0x4556454F. > Segmentation fault (core dumped) just fixed this in the CVS. thanks for the bug report. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Those who can't write, write manuals. |
From: Stig E S. <st...@ii...> - 2002-02-23 18:10:10
|
Quoting Sam Steingold (sd...@gn...): | > * In message <200...@ap...> | > * On the subject of "[clisp-list] REINITIALIZE-INSTANCE problems" | > * Sent on Thu, 21 Feb 2002 17:40:12 +0100 | > * Honorable Stig E Sandoe <st...@ii...> writes: | > | > *** - handle_fault error2 ! address = 0x4556454F not in | > [0x202A6000,0x2036220C) ! | > SIGSEGV cannot be cured. Fault address = 0x4556454F. | > Segmentation fault (core dumped) | | just fixed this in the CVS. good :-) | thanks for the bug report. do you have an estimate when this (imho important) fix comes in a released clisp? ie when will the next clisp be released, it has been quite awhile, and looking at clisp-devel it seems to me that current cvs contains many important bug-fixes. May I politely ask for "release early, release often"? :-) -- ------------------------------------------------------------------ Stig Erik Sandoe st...@ii... http://www.ii.uib.no/~stig/ |
From: Sam S. <sd...@gn...> - 2002-02-24 05:33:54
|
> * In message <200...@ap...> > * On the subject of "Re: [clisp-list] REINITIALIZE-INSTANCE problems" > * Sent on Sat, 23 Feb 2002 19:09:52 +0100 > * Honorable Stig E Sandoe <st...@ii...> writes: > > do you have an estimate when this (imho important) fix comes > in a released clisp? ie when will the next clisp be released, > it has been quite awhile, and looking at clisp-devel it seems > to me that current cvs contains many important bug-fixes. read CLISP/src/ChangeLog, find the bug you need, get the fixes you need from the CVS, apply the patches, build from the sources. > May I politely ask for "release early, release often"? :-) may I politely ask for money? :-) I have been unemployed for 3 weeks now, and if you can offer me a contract doing CLISP development, it would be nice. If you have a permanent position for me in the Boston, MA area, it would be even better (<http://www.podval.org/~sds/resume.html>). [just in case - CLISP is and always will be under GPL. don't worry] -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Trespassers will be shot. Survivors will be SHOT AGAIN! |
From: Todd S. <ts...@op...> - 2002-02-28 23:00:23
|
Sam Steingold <sd...@gn...> writes: > > * In message <200...@ap...> > > * On the subject of "Re: [clisp-list] REINITIALIZE-INSTANCE problems" > > * Sent on Sat, 23 Feb 2002 19:09:52 +0100 > > * Honorable Stig E Sandoe <st...@ii...> writes: > > > > do you have an estimate when this (imho important) fix comes > > in a released clisp? ie when will the next clisp be released, > > it has been quite awhile, and looking at clisp-devel it seems > > to me that current cvs contains many important bug-fixes. > > read CLISP/src/ChangeLog, find the bug you need, get the fixes you need > from the CVS, apply the patches, build from the sources. Are there things that you feel need to be done before the next release? In other words, are there things people can do (besides donate funding) to hasten the next release of clisp? Todd |
From: Sam S. <sd...@gn...> - 2002-03-01 04:53:39
|
> * In message <m36...@je...> > * On the subject of "Re: [clisp-list] REINITIALIZE-INSTANCE problems" > * Sent on 28 Feb 2002 18:00:15 -0500 > * Honorable Todd Sabin <ts...@op...> writes: > > Are there things that you feel need to be done before the next > release? I just committed the last (3rd) pre-test. All the brave, please do "cvs up" and try it out. I added weak hash-tables, which necessitated yet another fas-file format change - the last before the release (I wanted to get weak HTs into 2.28 because I want to avoid fas-file format changes for some time). Now, the things on the TODO list (FIUAR = "Forget It Until After Release"): 1. trace of flet/label locals (me, FIUAR) 2. Todd's patches (require Bruno's approval, FIUAR) 3. iconv compilation issues on some platforms (Bruno) 4. Solaris/gcc3/FFC failures (Bruno) 5. format/pplb, other pretty-print issues (me, FIUAR) 6. message translations; so far we have Stefan <li...@fr...> doing the German messages. We need French, Spanish and Dutch. Arseny, are you doing Russian? 7. libltdl integration for portability of dynamic modules, including woe32 (Bruno, FIUAR) 8. dirkey/ldap (me, FIUAR) 9. linux.lisp wait/waitpid patch needs to be tested and comitted... ... Bruno is very busy, so we will not wait for him (unless he speaks up and tells us to wait :-), even though the release quality will probably suffer. thus: as soon as the message translations are in, the release is out. you are _very_ welcome to pick up the aforementioned tasks, especially 5&8. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> I'm a Lisp variable -- bind me! |
From: Arseny S. <am...@ic...> - 2002-03-01 06:20:00
|
Hello, Sam, Friday, March 01, 2002, 2:49:20 PM, you wrote: Sam> Arseny, are you doing Russian? There is about 1/3 left (2/3 done). If translation detains the release, I'll speed translation up. -- Best regards, Arseny mailto:am...@ic... |
From: Sam S. <sd...@gn...> - 2002-03-02 02:41:26
|
> * In message <192...@ic...> > * On the subject of "Re[2]: [clisp-list] next release" > * Sent on Fri, 1 Mar 2002 16:18:57 +1000 > * Honorable Arseny Slobodjuck <am...@ic...> writes: > > Friday, March 01, 2002, 2:49:20 PM, you wrote: > > Sam> Arseny, are you doing Russian? > There is about 1/3 left (2/3 done). If translation detains the > release, I'll speed translation up. Arseny, I don't think you will finish it by 2.28. testing on win32 has much higher priority right now. Can you do it? (I do not have access to a woe32 machine) you need to build, do a 'make check', and try install.bat thanks. Russian translations will be added in 2.29. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Don't ascribe to malice what can be adequately explained by stupidity. |
From: Sam S. <sd...@gn...> - 2002-03-02 02:38:12
|
> * In message <m36...@je...> > * On the subject of "Re: [clisp-list] REINITIALIZE-INSTANCE problems" > * Sent on 28 Feb 2002 18:00:15 -0500 > * Honorable Todd Sabin <ts...@op...> writes: > > Are there things that you feel need to be done before the next > release? I want the release out by Monday. I tested it on Linux and FreeBSD. There will be _NO_ release until I get confirmation from the users that the current CVS CLISP compiles out of the box on at least win32 and Solaris. The translations can wait for the next release, but if I receive no build reports from Solaris and win32 people in the near future, I cannot make a release, thus the translations will get in &c. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Life is like a diaper -- short and loaded. |
From: Todd S. <ts...@op...> - 2002-03-02 05:38:08
|
Sam Steingold <sd...@gn...> writes: > > Are there things that you feel need to be done before the next > > release? > > I want the release out by Monday. > > I tested it on Linux and FreeBSD. Current CVS is failing for me on Redhat62, due to an ICONV_CONST mismatch. I think I've figured out the problem. unixconf.h.in has a line: #define ICONV_CONST but the autoconf sed stuff can't change it because there's no symbol it's defined to, which causes the sed expressions autoconf uses not to work. Apparently, the way to address that is by changing it thusly: diff -u -r1.30 unixconf.h.in --- src/unixconf.h.in 25 Dec 2001 17:48:28 -0000 1.30 +++ src/unixconf.h.in 2 Mar 2002 04:56:38 -0000 @@ -653,7 +653,7 @@ iconv_open, iconv, iconv_close functions. */ #undef HAVE_ICONV /* Define as const if the declaration of iconv() needs const. */ -#define ICONV_CONST +#undef ICONV_CONST /* OS services */ which matches the way all the other *_CONST stuff in there is done. And it does fix the build on Redhat 6.2. Redhat 7.2 also still works (and does without the patch). Unfortunately, with the patch the build doesn't work on FreeBSD 4.5 (I haven't tried 4.5 without the patch). The problem in that case is that it doesn't have ICONV at all, so no line is added to confdefs.h one way or the other about ICONV_CONST, which leaves the undef as is. So the compiler complains about ICONV_CONST being unrecognized. A possible way around _that_ is this: diff -u -r1.30 unixconf.h.in --- src/unixconf.h.in 25 Dec 2001 17:48:28 -0000 1.30 +++ src/unixconf.h.in 2 Mar 2002 05:18:32 -0000 @@ -653,8 +653,11 @@ iconv_open, iconv, iconv_close functions. */ #undef HAVE_ICONV /* Define as const if the declaration of iconv() needs const. */ -#define ICONV_CONST +#undef ICONV_CONST +#ifndef ICONV_CONST +#define ICONV_CONST +#endif /* OS services */ which makes sure that ICONV_CONST is defined as nothing, if not defined yet. Probably the right way is to use just the first patch, but also fix the code in configure to put a line in confdefs.h even if iconv isn't present, but it seems that configure itself is auto-generated somehow. I really have no idea how this autoconf stuff is supposed to work, in case you couldn't tell... :) Hopefully, someone else does... Anyway, with this second patch, Redhat 6.2 and 7.2 build for me fine. FreeBSD 4.5 manages to build the clisp binaries, but chokes on the Makefile in the suite directory, saying: LISP="`pwd`/lisp.run -M `pwd`/lispinit.mem -B `pwd` -N `pwd`/locale -Efile UTF-8 -norc"; export LISP; cd suite; make LISP="$LISP" "Makefile", line 2: Missing dependency operator "Makefile", line 5: Need an operator "Makefile", line 8: Need an operator make: fatal errors encountered -- cannot continue *** Error code 1 > There will be _NO_ release until I get confirmation from the users that > the current CVS CLISP compiles out of the box on at least win32 and > Solaris. I'll try the windows build tomorrow. Don't have any Solaris... Todd |
From: Arseny S. <am...@ic...> - 2002-03-02 11:13:45
Attachments:
spvw_language.diff
|
Hello Sam, Saturday, March 02, 2002, 12:37:02 PM, you wrote: Sam> Arseny, I don't think you will finish it by 2.28. Sam> testing on win32 has much higher priority right now. Sam> Can you do it? (I do not have access to a woe32 machine) Sam> you need to build, do a 'make check', and try install.bat Sam> thanks. It compiles fine except file spvw_language.d which needs declaration of uint. It seems to be declared in intparam.d -> intparam.h, which is not available in win32 (I think). So I suggest change uint to uintL - see patch. Install.bat and generated clisp.bat works fine. Tests are passed. Make check ok. CLHS doesn't work: [1]> (clhs 'car) *** - There is no package with name "NIL" 1. Break [2]> abort *** - There is no package with name "NIL" 1. Break [3]> abort Note: I tried to abort twice ! Maybe that will help [3]> (load "clhs.lisp") ;; Loading file clhs.lisp ... ;; Loading of file clhs.lisp is finished. T [5]> (clhs 'car) *** - There is no package with name "NIL" 1. Break [6]> up EVAL frame for form (PROGN (SYSTEM::%SET-PACKAGE-LOCK #:WOPL-1231 NIL) (SETQ SYSTEM::*CLHS-TABLE* (READ-FROM-FILE (SYSTEM::CLISP-DATA-FILE "clhs.txt") :OUT SYSTEM::OUT :PACKAGE "COMMON-LISP"))) (describe (code-char 1234)) works. I cannot makemake nor make distrib since I only have small subset of cygwin which moreover I didn't used much. I'll try to download and use it. Or I'll try to do distrib in FreeBSD (mount the FAT drive and make distrib with win32 makefile - is it valid approach ?). -- Best regards, Arseny mailto:am...@ic... |
From: Sam S. <sd...@gn...> - 2002-03-02 20:05:44
|
> * In message <144...@ic...> > * On the subject of "Re[4]: [clisp-list] next release" > * Sent on Sat, 2 Mar 2002 21:13:17 +1000 > * Honorable Arseny Slobodjuck <am...@ic...> writes: > > It compiles fine except file spvw_language.d which needs declaration > of uint. It seems to be declared in intparam.d -> intparam.h, which > is not available in win32 (I think). So I suggest change uint to uintL > - see patch. that was actually a typo. thanks. > *** - There is no package with name "NIL" fixed -- thanks! please try again. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> MS DOS: Keyboard not found. Press F1 to continue. |
From: Arseny S. <am...@ic...> - 2002-03-02 15:56:29
Attachments:
makemake.diff
|
Hello Sam, Saturday, March 02, 2002, 12:37:02 PM, you wrote: Sam> testing on win32 has much higher priority right now. I suggest that patch. The new data : section of makefile was totally unixoid and didn't work in win32. With that patch I was able to make distrib (from FreeBSD) which contained amongst other files data/UnicodeData.txt, data/clhs.txt, emacs/clisp-coding.el, emacs/clisp-indent.el, emacs/clisp-indent.lisp. -- Best regards, Arseny mailto:am...@ic... |
From: Sam S. <sd...@gn...> - 2002-03-02 19:42:19
|
Hi Arseny, > * In message <564...@ic...> > * On the subject of "Re[4]: [clisp-list] next release" > * Sent on Sun, 3 Mar 2002 01:55:36 +1000 > * Honorable Arseny Slobodjuck <am...@ic...> writes: > > I suggest that patch. The new data : section of makefile > was totally unixoid and didn't work in win32. > -echotab "cd data && \$(LN_S) ${PARENT_SRCTOPDIR}utils${NEXT}unicode${NEXT}ftp.unicode.org${NEXT}UnicodeData.txt UnicodeData.txt" > -echotab "cd data && \$(LN_S) ${PARENT_SRCTOPDIR}src${NEXT}clhs.txt clhs.txt" > +echotab "\$(LN_S) ${SRCTOPDIR_M}utils${NEXT_}unicode${NEXT_}ftp.unicode.org${NEXT_}UnicodeData.txt data${NEXT_}UnicodeData.txt" > +echotab "\$(LN_S) ${SRCTOPDIR_M}src${NEXT_}clhs.txt data${NEXT_}clhs.txt" this is broken when LN_S is "ln -s". please try the appended patch -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Oh Lord, give me the source code of the Universe and a good debugger! --- makemake.in.~1.225.~ Tue Feb 26 13:35:49 2002 +++ makemake.in Sat Mar 2 14:35:48 2002 @@ -2677,7 +2677,7 @@ DATA_FILES="UnicodeData.txt clhs.txt" echol "data :" -echotab "if test -d data; then rm -rf data; fi" +echotab "rm -rf data" echotab "mkdir data" echotab "cd data && \$(LN_S) ${PARENT_SRCTOPDIR}utils${NEXT}unicode${NEXT}ftp.unicode.org${NEXT}UnicodeData.txt UnicodeData.txt" echotab "cd data && \$(LN_S) ${PARENT_SRCTOPDIR}src${NEXT}clhs.txt clhs.txt" |
From: Sam S. <sd...@gn...> - 2002-03-02 23:39:27
|
> * In message <m3e...@je...> > * On the subject of "Re: [clisp-list] REINITIALIZE-INSTANCE problems" > * Sent on 02 Mar 2002 00:37:59 -0500 > * Honorable Todd Sabin <ts...@op...> writes: > > Current CVS is failing for me on Redhat62, due to an ICONV_CONST > mismatch. I think I've figured out the problem. unixconf.h.in > has a line: > > #define ICONV_CONST I _think_ I just fixed this, hopefully forever. Please _do_ test it on all your platforms. And I do not mean just you, Todd. Everyone who has been saying "release early; release often" - please do "cvs up" and build CLISP on your platform. You have nothing to lose: if you succeed, you will have the next stable version of CLISP a couple of days before everyone else, if you fail and report your problem, the next CLISP version will work on your platform out of the box, so you will not have to wait until the next release. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Do not worry about which side your bread is buttered on: you eat BOTH sides. |
From: Arseny S. <am...@ic...> - 2002-03-03 00:41:42
|
Hello Sam, Sunday, March 03, 2002, 5:37:52 AM, you wrote: >> I suggest that patch. The new data : section of makefile >> was totally unixoid and didn't work in win32. >> -echotab "cd data && \$(LN_S) ${PARENT_SRCTOPDIR}utils${NEXT}unicode${NEXT}ftp.unicode.org${NEXT}UnicodeData.txt UnicodeData.txt" >> -echotab "cd data && \$(LN_S) ${PARENT_SRCTOPDIR}src${NEXT}clhs.txt clhs.txt" >> +echotab "\$(LN_S) ${SRCTOPDIR_M}utils${NEXT_}unicode${NEXT_}ftp.unicode.org${NEXT_}UnicodeData.txt data${NEXT_}UnicodeData.txt" >> +echotab "\$(LN_S) ${SRCTOPDIR_M}src${NEXT_}clhs.txt data${NEXT_}clhs.txt" Sam> this is broken when LN_S is "ln -s". Sam> please try the appended patch Sam> DATA_FILES="UnicodeData.txt clhs.txt" Sam> echol "data :" Sam>-echotab "if test -d data; then rm -rf data; fi" Sam>+echotab "rm -rf data" Then, maybe, "-rm -rf data" ? Sam> echotab "mkdir data" Sam> echotab "cd data && \$(LN_S) ${PARENT_SRCTOPDIR}utils${NEXT}unicode${NEXT}ftp.unicode.org${NEXT}UnicodeData.txt I tried a lot of ways. In win32 LN_S=copy , which doesn't understand forward slashes on NT (it counts it as option). Since we use sed and rm (I personally use set of GNU file utilities ported for NT, must have for developer) we can use LN_S=cp. Or we need ${PARENT_SRCTOPDIR_} in addition to ${PARENT_SRCTOPDIR}. But rm still remains. To remove it we need $(RMRF1)=del /q /s $(RMRF2)=rmdir /q /s on win32 and $(RMRF1)=rm -rf $(RMRF2)= on unix and use both it every time or something similar (bat file rmrf.bat for example). Brrr... Using rm is less painful. BTW, there is ln port, it even gets rid of NT hardlinks. It seems to me that Makefile was revised to remove use of such utilities for win32 since last year, is it ? Another problem is data dependency. VC nmake looks to data directory creation date and does nothing if it exists (but files in it may be obsolete). I don't know whether it is a big problem. clhs working now. -- Best regards, Arseny mailto:am...@ic... |
From: Sam S. <sd...@gn...> - 2002-03-03 06:36:24
|
Hi Arseny, > * In message <972...@ic...> > * On the subject of "Re[6]: [clisp-list] next release" > * Sent on Sun, 3 Mar 2002 10:40:43 +1000 > * Honorable Arseny Slobodjuck <am...@ic...> writes: > > It seems to me that Makefile was revised to remove use of such > utilities for win32 since last year, is it ? i do not recall it. when? what ChangeLog entry? CVS revision? > Another problem is data dependency. VC nmake looks to data directory > creation date and does nothing if it exists (but files in it may be > obsolete). I don't know whether it is a big problem. this is not a big deal. > clhs working now. good. I _think_ I fixed the problems you mention in the makemake.in in the CVS. please regenerate your makefile and try again. thanks. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Those who can laugh at themselves will never cease to be amused. |
From: Todd S. <ts...@op...> - 2002-03-03 02:31:51
|
Sam Steingold <sd...@gn...> writes: > > Current CVS is failing for me on Redhat62, due to an ICONV_CONST > > mismatch. I think I've figured out the problem. unixconf.h.in > > has a line: > > > > #define ICONV_CONST > > I _think_ I just fixed this, hopefully forever. > Please _do_ test it on all your platforms. Works on Redhat 6.2 and 7.2. FreeBSD 4.5 builds to the point where it complains about the Makefile in suite, as mentioned in the last mail. Is that normal? I tried win32 with VS6 earlier today. Builds with a couple minor problems fixed. The first is that del fails on non-existant files, so --- ../win32msvc/makefile.msvc5 Mon Nov 19 14:38:15 2001 +++ makefile Sat Mar 2 16:40:50 2002 @@ -50,10 +50,10 @@ MSVCDIR = "C:/Program Files/Microsoft Visual Studio/VC98" MAKE = nmake -RM = del +RM = -del CP = copy LN_S = copy MV = ren The other is that the supplied install.bat is referencing src/install.lisp, but appears to want to be run from src itself, which doesn't work. RCS file: /cvsroot/clisp/clisp/src/install.bat,v retrieving revision 1.2 diff -u -b -r1.2 install.bat --- install.bat 9 Jun 2001 17:03:13 -0000 1.2 +++ install.bat 3 Mar 2002 02:11:53 -0000 @@ -6,6 +6,6 @@ echo press C-c to abort pause -lisp.exe -B . -M lispinit.mem -norc -C src/install.lisp +lisp.exe -B . -M lispinit.mem -norc -C install.lisp pause Todd |
From: Sam S. <sd...@gn...> - 2002-03-03 05:53:08
|
> * In message <m3a...@je...> > * On the subject of "Re: [clisp-list] REINITIALIZE-INSTANCE problems" > * Sent on 02 Mar 2002 21:31:43 -0500 > * Honorable Todd Sabin <ts...@op...> writes: > > FreeBSD 4.5 builds to the point where it complains about the Makefile > in suite, as mentioned in the last mail. Is that normal? I just fixed this. unfortunately this required removing the gnu-make-specific IF construct (is there a portable on across all make implementations? :-) this means that cygwin users will have to edit the Makefile to run the testsuite or specify the LISP make var on the command line. > I tried win32 with VS6 earlier today. Builds with a couple minor > problems fixed. The first is that del fails on non-existant files, > so > > --- ../win32msvc/makefile.msvc5 Mon Nov 19 14:38:15 2001 > +++ makefile Sat Mar 2 16:40:50 2002 > @@ -50,10 +50,10 @@ > MSVCDIR = "C:/Program Files/Microsoft Visual Studio/VC98" > > MAKE = nmake > -RM = del > +RM = -del > CP = copy > LN_S = copy > MV = ren please regenerate your makefile.msvc5 using make -f Makefile.devel win32msvc/makefile.msvc5 > The other is that the supplied install.bat is referencing > src/install.lisp, but appears to want to be run from src itself, which > doesn't work. you are missing the point. install.bat is for the users of the binary distributions. a binary distribution has lisp.exe and lispinit.mem at the top-level and install.lisp in the src subdirectory. you should use cygwin to create a binary distribution $ make distrib and then install it somewhere and run install.bat there. -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> Why use Windows, when there are Doors? |
From: Arseny S. <am...@ic...> - 2002-03-03 04:36:06
|
Hello Todd, Sunday, March 03, 2002, 12:31:43 PM, you wrote: Todd> MAKE = nmake Todd> -RM = del Todd> +RM = -del Always in makemake.in, Makefile.msvc5 just needs to be recreated. Todd> The other is that the supplied install.bat is referencing Todd> src/install.lisp, but appears to want to be run from src itself, which Todd> doesn't work. install.bat intended to be called from distribution, where src/ and data/ (which also is in makefile.in) exists. -- Best regards, Arseny mailto:am...@ic... |
From: Arseny S. <am...@ic...> - 2002-03-03 10:16:41
|
Hello Sam, Sunday, March 03, 2002, 4:31:56 PM, you wrote: >> It seems to me that Makefile was revised to remove use of such >> utilities for win32 since last year, is it ? Sam> i do not recall it. Sam> when? Sam> what ChangeLog entry? Sam> CVS revision? I remember that when I tried compile clisp on win32 for the first time, it was impossible to compile without cp, sed and uniq. I didn't tried recently w/o utilities, but at least cp was changed to copy. Uniq was removed at all. Sed is called on distrib stage only. In last patch you removed call to rm and added del instead. That's what I mean. Sam> I _think_ I fixed the problems you mention in the makemake.in in the Sam> CVS. Sam> please regenerate your makefile and try again. Genialno. Works. Del asks permission to delete files by pattern, so it needs the /q switch: Index: makemake.in =================================================================== RCS file: /cvsroot/clisp/clisp/src/makemake.in,v retrieving revision 1.226 diff -u -r1.226 makemake.in --- makemake.in 3 Mar 2002 06:34:17 -0000 1.226 +++ makemake.in 3 Mar 2002 09:11:42 -0000 @@ -612,7 +612,7 @@ # RM = command for deleting files if [ $H_DOS = "true" ] ; then - RM='del' + RM='del /q' else if [ $HOS = acorn ] ; then RM='remove' Problems with make check (I didn't make check with newly generated makefile until now): francais spanish comp.exe returns error on compare (because of gensyms). I never went further, but now when I changing comp to -comp, testsuite doesn't work: copy doesn't take arbitrary number of args: mkdir suite cd suite && copy ..\..\tests\Makefile ..\..\tests\*.lisp ..\..\tests\*.tst . Recently I created separate makefile for tests and it was passed. This is not vitally important though. WARNING: INTERN("STOREFORM"): #<PACKAGE LDAP> is locked Ignore the lock and proceed Is that normal ? A lot of such warnings prints out during make check. -- Best regards, Arseny mailto:am...@ic... |
From: Sam S. <sd...@gn...> - 2002-03-03 17:51:01
|
Hi Arseny > * In message <543...@ic...> > * On the subject of "Re[8]: [clisp-list] next release" > * Sent on Sun, 3 Mar 2002 20:15:36 +1000 > * Honorable Arseny Slobodjuck <am...@ic...> writes: > > - RM='del' > + RM='del /q' done > Problems with make check (I didn't make check with newly generated > makefile until now): > francais > spanish > comp.exe returns error on compare (because of gensyms). that's okay. > I never went further, but now when I changing comp to > -comp, testsuite doesn't work: copy doesn't take arbitrary > number of args: fixed. > WARNING: > INTERN("STOREFORM"): #<PACKAGE LDAP> is locked > Ignore the lock and proceed this is normal when running "make test" -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> MS Windows: error: the operation completed successfully. |
From: Eric M. <ema...@la...> - 2002-03-03 17:39:18
|
>>>>> "Sam" == Sam Steingold <sd...@gn...> writes: Sam> I tested it on Linux and FreeBSD. on Linux, I get an error during "make install". I build in a private directory (last argument to configure is the directory /tmp/clisp-gcc). Build and tests work fine (including modules wildcard and regexp and --export-syscalls), then in the /tmp/clisp-gcc directory, ,---- | % sudo make install | test -d /usr/local || mkdir /usr/local | test -d /usr/local || mkdir /usr/local | test -d /usr/local/lib || mkdir /usr/local/lib | test -d /usr/local/lib/clisp || mkdir /usr/local/lib/clisp | test -d /usr/local/lib/clisp/data || mkdir /usr/local/lib/clisp/data | /usr/bin/install -c -m 644 data/UnicodeData.txt /usr/local/lib/clisp/data/UnicodeData.txt | /usr/bin/install: cannot stat `data/UnicodeData.txt': No such file or directory | make: *** [install-bin] Error 1 | % ls -l data | total 0 | 0 lrwxrwxrwx 1 emarsden tsf 73 Mar 3 18:24 UnicodeData.txt -> ..//opt/src/cvs-clisp/clisp/utils/unicode/ftp.unicode.org/UnicodeData.txt | 0 lrwxrwxrwx 1 emarsden tsf 40 Mar 3 18:24 clhs.txt -> ..//opt/src/cvs-clisp/clisp/src/clhs.txt | % `---- I ran configure from the directory /opt/src/cvs-clisp/clisp. Sam> There will be _NO_ release until I get confirmation from the users that Sam> the current CVS CLISP compiles out of the box on at least win32 and Sam> Solaris. is the FFI supposed to work on Solaris? I generally build without it, but I just tried to build with the regexp and wildcard modules, and it fails with ,---- | /tmp/clisp-gcc/lisp.run -M /tmp/clisp-gcc/lispinit.mem -B /tmp/clisp-gcc -N /tmp/clisp-gcc/locale -Efile UTF-8 -norc -q -c regexp.lisp | Compiling file /tmp/clisp-gcc/regexp/regexp.lisp ... | *** - There is no package with name "FFI" `---- This is on SunOS 5.7 with gcc 2.95.2. The avcall stuff seems to have compiled ok: ,---- | /tmp/clisp-gcc$ cat avcall/minitests.output.sparc-sun-solaris2.7 | Int f(Int,Int,Int):({1},{2},{3})->{6} | Int f(Int,Int,Int):({1},{2},{3})->{9} | J f(J,int,J):({47,11},2,{73,55})->{120,68} | J f(J,int,J):({47,11},2,{73,55})->{9,902231815} `---- -- Eric Marsden <URL:http://www.laas.fr/~emarsden/> |
From: Sam S. <sd...@gn...> - 2002-03-04 02:26:32
|
> * In message <wzi...@me...> > * On the subject of "Re: [clisp-list] REINITIALIZE-INSTANCE problems" > * Sent on Sun, 03 Mar 2002 18:39:04 +0100 > * Honorable Eric Marsden <ema...@la...> writes: > > >>>>> "Sam" == Sam Steingold <sd...@gn...> writes: > > Sam> I tested it on Linux and FreeBSD. > > | % ls -l data > | total 0 > | 0 lrwxrwxrwx 1 emarsden tsf 73 Mar 3 18:24 UnicodeData.txt -> ..//opt/src/cvs-clisp/clisp/utils/unicode/ftp.unicode.org/UnicodeData.txt > | 0 lrwxrwxrwx 1 emarsden tsf 40 Mar 3 18:24 clhs.txt -> ..//opt/src/cvs-clisp/clisp/src/clhs.txt okay - fixed, please try again. (this "try again", unfortunately, includes those of you, who have already tested the latest CVS CLISP and reported success. sorry guys. )-: > Sam> There will be _NO_ release until I get confirmation from the users that > Sam> the current CVS CLISP compiles out of the box on at least win32 and > Sam> Solaris. > > is the FFI supposed to work on Solaris? yes. > I generally build without it, but I just tried to build with the > regexp and wildcard modules, and it fails with > > ,---- > | /tmp/clisp-gcc/lisp.run -M /tmp/clisp-gcc/lispinit.mem -B /tmp/clisp-gcc -N /tmp/clisp-gcc/locale -Efile UTF-8 -norc -q -c regexp.lisp > | Compiling file /tmp/clisp-gcc/regexp/regexp.lisp ... > | *** - There is no package with name "FFI" > `---- try $ ./configure --with-dynamic-ffi -- Sam Steingold (http://www.podval.org/~sds) running RedHat7.2 GNU/Linux Keep Jerusalem united! <http://www.onejerusalem.org/Petition.asp> Read, think and remember! <http://www.iris.org.il> <http://www.memri.org/> A man paints with his brains and not with his hands. |
From: Eric M. <ema...@la...> - 2002-03-04 09:09:32
|
>>>>> "Sam" =3D=3D Sam Steingold <sd...@gn...> writes: Sam> okay - fixed, please try again. Sam> (this "try again", unfortunately, includes those of you, who have Sam> already tested the latest CVS CLISP and reported success. Sam> sorry guys. )-: right, it works now, thanks. =20 =20 ecm> is the FFI supposed to work on Solaris?=20 Sam> yes. Sam> try Sam> $ ./configure --with-dynamic-ffi thanks, is there a reason why this defaults to on for linux/x86, but needs to be specified explicitly on Solaris? Also, it would be nice if it appeared in the output of =ABconfigure --help=BB. Apparently the avcall tests fail: ,---- | /tmp/clisp-gcc$ more avcall/minitests.output.sparc-sun-solaris2.7=20 | Int f(Int,Int,Int):({1},{2},{3})->{6} | Int f(Int,Int,Int):({1},{2},{3})->{9} | J f(J,int,J):({47,11},2,{73,55})->{120,68} | J f(J,int,J):({47,11},2,{73,55})->{9,902231815} `---- I'm afraid that the error in avcall-sparc.s isn't immediately evident to me :-) =20 --=20 Eric Marsden <URL:http://www.laas.fr/~emarsden/> |