[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: LaTeX conversion

> Ok, I've got it now.  There seems to be a problem with some of the chapter
> menus (at least under emacs info.)  Many of the chapter menus seem to be
> missing.  For example, the debugger, compiler and advanced compiler
> chapters.   You can jump into the text in other ways, but there is no menu
> in the chapter node.
Yes, it was just a first cut; I really hadn't put the menus in fully, as
I was working on the part of LaTeXinfo that does that at the time.

> Also, for some reason if you try to select the serve-event chapter in the
> top-level menu, you get an error that that node doesn't exist.  I note in
> completing at the goto prompt, that it refuses to complete past the #\: in
> SYSTEM:SERVE-EVENT.  Perhaps colons in menu items are a no-no?

Yes, Info doesn't like colons.  Also, the section names have to be
unique, so I've had to make some minor changes to the sources to get
everything a unique header.  Plus a few changes where \deffun ..
\enddefmac gets caught by LaTeX, and a couple of other bizarities that
are legal in Scribe but are too painful to try and catch with the
regexps that convert Scribe to LaTeXinfo.  I'll give you the diffs to
check in on the sources in a couple of days; with these checked in,
the Scribe->LaTeXinfo conversion should be automatic, other than
manually fixing up the titlepage.

I've made some progress on incorporating Steele's macros; I'll finish them
off, and bring the documentation in line with the style.  I must say,
it is nice to work in cmulisp on a system with the manual on line.

I've got a number of questions for you on cmulisp.

Is anyone there interested in graphical user interfaces for cmulisp?
I'm try to port Berkeley's Picasso to cmulisp.  It's running under Allegro
4.0 with CLOS, so I need to at least upgrade the cmu pcl to Rev 4 PCL.
Has anyone brought CMU's PCL up closer to CLOS?  Or the CLX to 5.0?

I did a diff of REV 2 PCL (the one cmulisp uses) on arisia.xerox.com
with the REV 4 PCL, and applied the diffs to the CMU PCL, with no
significant problems.  (Let me know if you want the final diffs).  I
wanted to build a new lisp.core using the new PCL, but I seem to get
caught in a bootstraping loop:

>    3) The file ldb/lisp.h is missing from the source tar.  Hence the binary
>    cannot be built from the sources without the binaries.
> That's a bit of a hack (bootstrapping issue.)  You need to either run
> GENESIS with no input files to generate lisp.h, or copy the generated file
> from e.g. /afs/cs/project/clisp/sun4c_41/beta/ldb/lisp.h

I picked up that file a while ago, but it doesn't make it:

font%mike 217>> make
cc   -target sun4 -c  ldb.c
"ldb.c", line 136: PSEUDO_ATOMIC_ATOMIC undefined
"ldb.c", line 137: PSEUDO_ATOMIC_INTERRUPTED undefined
*** Error code 1
make: Fatal error: Command failed for target `ldb.o'

Looks like you need a ldb.map as well, to run lisp::genesis, but I can't 
pick up an ldb.map now:

Name (lisp-rt1.slisp.cs.cmu.edu:mike): anonymous
331 Guest login ok, send username@node as password.
230 Filenames can not have '/..' in them.
ftp> cd /afs/cs/project/clisp/sun4c_41/beta/ldb
530 Access not allowed for guest users for path /afs/cs/project/clisp/sun4c_41/beta/ldb

Finally I did a nm -gp of the binary distribution executable, and
assumed that that was my ldb.  That seems to work, but it might an idea
to distribute the lisp.h and ldb.map with the binary or source
distributions. Then people can bootstrap themselves with what's distributed.

The file lib/site-init.lisp should probably have a #-no-hemlock in front of
the setv's.

Is there any way of cat'ing fasl files together?  Allegro has a nice feature
to it where cat *.fasl > BIG.fasl results in a loadable fasl file.  So if
I'm trying to build an image, I can add or leave out things easily, and the
BIG.fasl files are fast enough to load that I don't need to do a save-lisp
except for programs I use constantly.  WIth the size of the lisp.core files,
it makes it quite difficult if you have to keep multiple core files around.

Is this intrinsically difficult to do in cmulisp?  Would a simple awk script
or C program suffice to combine the relevant information from many fasl files
into one?  If so, it might make things easier for building packages such as
PCL, Hemlock, CLX,  Picasso etc.

Many thanks in advanceeee, and I should have things ready for you at the
beginning of next week.


Mike Clarkson					mike@apl.ists.ca
C.R.E.S.S.					      - or -
York University, North York, Ontario,		mike@cmm.ensmp.fr
CANADA M3J 1P3					+33-1-64 69 47 98