[Ecls-list] how to build Ecl with preset higher memory bounds

Oliver Kullmann O.Kullmann at swansea.ac.uk
Sat Jan 22 19:14:28 UTC 2011


Hi,

Juan Jose Garcia-Ripoll <juanjose.garciaripoll at go...> - 2011-01-21 13:35
wrote:

I think you are looking for this:

http://ecls.sourceforge.net/new-manual/ch20s03.html

The C equivalent

http://ecls.sourceforge.net/new-manual/re41.html

But the limits you are thinking about are, in my opinion, too large.
Specially frame_stack_ecl. Are you going to have 2^14 NESTED
tagbody/block/catch/unwind-protect structures?

-----------------------

The information http://ecls.sourceforge.net/new-manual/ch20s03.html
I'm using already, but it doesn't work: It works from the (Maxima) command-line
but not as an automated process, since then always some Lisp-process is already
working, and Maxima segfaults --- and these batch-processes are the core of
our activity.

Thus I just want to get rid off the problem once and for all, and just build
Ecl with our bounds built-in.

I don't understand what http://ecls.sourceforge.net/new-manual/re41.html
refers to.

In my understanding 5 build-options for Ecl for setting the default-values of those
five bounds would be appropriate for everybody.
But if this is not possible, we can just change the source-code of Ecl --- we only need to
know which files to change.

The web-page mentions ~/.eclrc: I don't know whether Maxima uses that, and
in any case we can't use it, since it interferes with the user-space, and our
platform must not interfere in any way with user-data (we install everything
locally, and must keep all data local).

Just building Ecl in the way we want it would get rid off all these problems.

Regarding the values: I think that all of these values come from experience, where
we got an error message mentioning this bound, and after increasing the bound
we got it working. Though I'm not fully sure; once we get the larger bounds working
I'll try to decrease it (yet we just can't run all the processes which require
large memory, since we can't set the bounds for the batch-processes).

Oliver

P.S. I copied that e-mail from the archive-web-page --- apparently I didn't get it.
Strange that some e-mails apparently don't get through (while others do).





More information about the ecl-devel mailing list