Embeddable Common-Lisp

On Tuesday 27 May 2003 13:00, Dave Pearson wrote:
> Is it by design that ecl's `format' treats ~C in the way all other CL
> implementations seem to treat ~:C? [...]
> whereas with ecl I get:
> | > (format nil "~C" #\Space)
> |
> | "Space"

Thanks for reporting this bug. I have fixed it once more, and updated the 
patch-0.9-current.gz for it.

Best regards

Juanjo

* Juan Jose Garcia Ripoll <worm@...> [2003-05-27 14:40:56 +0200]:

> On Tuesday 27 May 2003 13:00, Dave Pearson wrote:
> > Is it by design that ecl's `format' treats ~C in the way all other CL
> > implementations seem to treat ~:C? [...]
> > whereas with ecl I get:
> > | > (format nil "~C" #\Space)
> > |
> > | "Space"
> 
> Thanks for reporting this bug. I have fixed it once more, and updated the
> patch-0.9-current.gz for it.

Thanks for that. BTW, I don't know if this is important or not but that new
patch-0.9-current complains about a small number of missing files when
applied against the stock 0.9 tarball sources. It seems to be a CHANGELOG in
the src directory and some items pertaining to format tests (so I'm guessing
they're not important).

-- 
Dave Pearson
http://www.davep.org/

* Dave Pearson <davep@...> [2003-05-28 09:24:55 +0100]:

> Thanks for that. BTW, I don't know if this is important or not but that
> new patch-0.9-current complains about a small number of missing files when
> applied against the stock 0.9 tarball sources. It seems to be a CHANGELOG
> in the src directory and some items pertaining to format tests (so I'm
> guessing they're not important).

Unarchiving a clean 0.9 and applying the new patch seems to stop ecl from
building with this error:

,----
| gcc -c -I../h -I/usr/local/src/ecl-0.9/src/c -I/usr/local/src/ecl-0.9/src/h -I/usr/local/src/ecl-0.9/src/gc/include -g -O2 -fstrict-aliasing -Dlinux -o file.o file.c
| /usr/local/src/ecl-0.9/src/c/file.d: In function `flisten':
| /usr/local/src/ecl-0.9/src/c/file.d:1007: variable `tv' has initializer but incomplete type
| /usr/local/src/ecl-0.9/src/c/file.d:1007: warning: excess elements in struct initializer after `tv'
| /usr/local/src/ecl-0.9/src/c/file.d:1007: warning: excess elements in struct initializer after `tv'
| /usr/local/src/ecl-0.9/src/c/file.d:1007: storage size of `tv' isn't known
| make[2]: *** [file.o] Error 1
| rm list.c file.c disassembler.c main.c eval.c character.c gfun.c compiler.c reference.c clos.c symbol.c package.c interpreter.c instance.c
| make[2]: Leaving directory `/usr/local/src/ecl-0.9/build/c'
| make[1]: *** [libecl.a] Error 2
| make[1]: Leaving directory `/usr/local/src/ecl-0.9/build'
| make: *** [all] Error 2
`----

This is on RedHat GNU/Linux 6.2, patched up, GCC 2.91 with ecl configured
with the command "./configure --prefix=/usr/local/stow/ecl --with-tcp".

-- 
Dave Pearson:                   |     lbdb.el - LBDB interface.
http://www.davep.org/           |  sawfish.el - Sawfish mode.
Emacs:                          |  uptimes.el - Record emacs uptimes.
http://www.davep.org/emacs/     | quickurl.el - Recall lists of URLs.

On Wednesday 28 May 2003 10:24, Dave Pearson wrote:
> Thanks for that. BTW, I don't know if this is important or not but that new
> patch-0.9-current complains about a small number of missing files when
> applied against the stock 0.9 tarball sources. It seems to be a CHANGELOG
> in the src directory and some items pertaining to format tests (so I'm
> guessing they're not important).

Yes, I forgot to patch the patch file manually :-) Seems that some files were 
missing in the distribution: CHANGELOG, which is just documentation, and the 
test files, which have to be distributed separately due to license issues.

BTW, did the system build properly?

Best regards,

Juanjo

* Juan Jose Garcia Ripoll <worm@...> [2003-05-28 15:09:18 +0200]:

> BTW, did the system build properly?

Not for me, see my email <20030528101922.GB5924@...> from
yesterday.

-- 
Dave Pearson
http://www.davep.org/

On Wednesday 28 May 2003 12:19, Dave Pearson wrote:
> Unarchiving a clean 0.9 and applying the new patch seems to stop ecl from
> building with this error: [...]

It's my fault. I forgot to include a file header (sys/select.h) which might be 
required in some systems. I have added an updated patch to 
http://www.sf.net/projects/ecls (under Releases)

Sorry it took so long to fix.

Juanjo


-- 
Max-Planck-Institut fuer Quantenoptik	+49/(0)89/32905-345
Hans-Kopfermann-Str. 1, D-85748	 http://www.mpq.mpg.de/Theorygroup/CIRAC/
Garching b. Muenchen, Germany		Juan.Ripoll@...

* Juan Jose Garcia Ripoll <juan.ripoll@...> [2003-06-02 19:31:03 +0200]:

> On Wednesday 28 May 2003 12:19, Dave Pearson wrote:
>
> > Unarchiving a clean 0.9 and applying the new patch seems to stop ecl from
> > building with this error: [...]
> 
> It's my fault. I forgot to include a file header (sys/select.h) which
> might be required in some systems. I have added an updated patch to
> http://www.sf.net/projects/ecls (under Releases)

Thanks, I'll grab that and have a play.

> Sorry it took so long to fix.

No worries.

-- 
Dave Pearson
http://www.davep.org/

* Dave Pearson <davep@...> [2003-06-03 08:57:56 +0100]:

> * Juan Jose Garcia Ripoll <juan.ripoll@...> [2003-06-02 19:31:03 +0200]:
> 
> > It's my fault. I forgot to include a file header (sys/select.h) which
> > might be required in some systems. I have added an updated patch to
> > http://www.sf.net/projects/ecls (under Releases)
> 
> Thanks, I'll grab that and have a play.

I untarred a fresh 0.9, applied the patch (which applied without any
problems) configured and built and the build process stopped with the same
problem:

-- cut here ----------------------------------------------------------------
gcc -c -I../h -I/usr/local/src/ecl-0.9/src/c -I/usr/local/src/ecl-0.9/src/h -I/usr/local/src/ecl-0.9/src/gc/include -g -O2 -fstrict-aliasing -Dlinux -o file.o file.c
/usr/local/src/ecl-0.9/src/c/file.d: In function `flisten':
/usr/local/src/ecl-0.9/src/c/file.d:1009: variable `tv' has initializer but incomplete type
/usr/local/src/ecl-0.9/src/c/file.d:1009: warning: excess elements in struct initializer after `tv'
/usr/local/src/ecl-0.9/src/c/file.d:1009: warning: excess elements in struct initializer after `tv'
/usr/local/src/ecl-0.9/src/c/file.d:1009: storage size of `tv' isn't known
make[2]: *** [file.o] Error 1
rm list.c file.c disassembler.c main.c eval.c character.c gfun.c compiler.c reference.c clos.c symbol.c package.c interpreter.c instance.c
make[2]: Leaving directory `/usr/local/src/ecl-0.9/build/c'
make[1]: *** [libecl.a] Error 2
make[1]: Leaving directory `/usr/local/src/ecl-0.9/build'
make: *** [all] Error 2
-- cut here ----------------------------------------------------------------

-- 
Dave Pearson
http://www.davep.org/

What system are you compiling ECL on? Can you please, after
#include <sys/select.h>
add
#include <sys/time.h>
#include <sys/types.h>
#include <unistd.h>
If this compiles, please tell me, because the officially required header 
(according to  POSIX 1003. etc) is sys/select.h, but it seems that your 
system requires also other headers for the "timeval" structure.

Regards,

Juanjo

-- 
Max-Planck-Institut fuer Quantenoptik	+49/(0)89/32905-345
Hans-Kopfermann-Str. 1, D-85748	 http://www.mpq.mpg.de/Theorygroup/CIRAC/
Garching b. Muenchen, Germany		Juan.Ripoll@...

* Juan Jose Garcia Ripoll <juan.ripoll@...> [2003-06-03 13:46:50 +0200]:

> What system are you compiling ECL on? 

,----[ From <20030528101922.GB5924@...> ]
| This is on RedHat GNU/Linux 6.2, patched up, GCC 2.91 with ecl configured
| with the command "./configure --prefix=/usr/local/stow/ecl --with-tcp".
`----

>                                       Can you please, after
> #include <sys/select.h>
> add
> #include <sys/time.h>
> #include <sys/types.h>
> #include <unistd.h>
> If this compiles, please tell me, because the officially required header 
> (according to  POSIX 1003. etc) is sys/select.h, 

The above lines seemed to do the trick for file.d. However, the build now
stops with this:

,----
| make[2]: *** No rule to make target `ffi.o', needed by `../libecl.a'.  Stop.
`----

>                                                  but it seems that your
> system requires also other headers for the "timeval" structure.

Yes, on this (admittedly aged) system it is in <bits/time.h> which seems to
have been included from <sys/time.h>.

-- 
Dave Pearson
http://www.davep.org/