autoloading messages, was: Maxima 5.12.0 rc1 in cvs & SF download

Volker,

Okay, if I understand the problem correctly, you're trying to set the
autoload of printf using the correct full pathname, but at the time of
compilation, when this expression is evaluated, file_search_lisp is not yet
known, and your call to file_search therefore fails.

It seems to me that it is not possible to set autoload to a function call
that is evaluated at runtime. Setting autoload to "stringproc" with no path
produces another error: Maxima complains that stringproc is not found in
file_search_maxima.

However, this suggests a solution! Well, two solutions, as a matter of fact,
but the obvious one doesn't work: setting autoload to "stringproc.lisp"
doesn't do the trick. But setting it to "stringproc" does, so long as you
add a new file to the stringproc directory named stringproc.mac, containing
just the single line,

load("stringproc.lisp");

as this allows Maxima to find "stringproc" (i.e., stringproc.mac) using
default search settings.


To sum up, three changes are needed to make all this work:

1. Use

#-gcl (setf (get '$printf 'autoload) "stringproc")

in max_ext.lisp;

2. Create stringproc.mac in share/contrib/stringproc; and

3. Make sure stringproc.mac is installed by make install.


I don't know if this is the "right" way of doing it, as I am just hacking my
way through the autoloading mechanism, but it does appear to work. I've
tested it with SBCL/CLISP/CMUCL using maxima-local; I also installed Maxima
with SBCL/CLISP on a clean test virtual machine, and tested that the
mechanism works in the installed configuration as well.

If you wish, I can make and commit the necessary changes, just let me know
what you prefer.


Viktor



 

-----Original Message-----
From: van Nek [mailto:van.nek at arcor.de] 
Sent: Monday, April 16, 2007 11:39 AM
To: Viktor T. Toth
Cc: Maxima at math.utexas.edu
Subject: Re: [Maxima] autoloading messages, was: Maxima 5.12.0 rc1 in cvs &
SF download

Am 15 Apr 2007 um 17:16 hat Viktor T. Toth geschrieben:

> > But before I completely undo my changes please give them
> > one more try by commenting out line 176
> 
> Done that, and it compiles cleanly. 

Good news!

What gets broken when this line is
> removed?

This line is independend from the others and if it is removed, this doesn't
break anything. 
This just means, that at the moment I do not find a way to autoload the
macro function 
$printf from stringproc.lisp for any other Lisp than GCL. Ideas?

Viktor, thanks for testing my commits on Linux
Best
Volker