Interest on ECL port?

[Juanjo]

It seems that Mudballs is more repeatable and better organized than
ASDF, including support for the latter. I was wondering whether
Mudballs supports or wishes to support ECL (http://
ecls.sourceforge.net)

In ECL we currently support building shared libraries and programs out
of ASDF files. It is really simple: get a list of the files to be
compiled from ASDF and compile and link them against the ECL library.
The process is automated via several classes in ECL and it only has
one drawback: ASDF is so badly designed that it may decided to load
and compile the same file several times and in an arbitrary order.

I was wondering whether the same thing could be done in a more
reasonable way using Mudballs. As I said, the goal is to generate a
list of operations to be performed: load, compile, etc, but 1) the
compilation steps will generate not a FASL but rather an object file
and 2) a final step linking all object files should be performed.

In case people are interested, I would be glad to help.

Juanjo

[Sean Ross]

On 25 Mar 2009, at 09:29, Juanjo wrote:

>
> It seems that Mudballs is more repeatable and better organized than
> ASDF, including support for the latter. I was wondering whether
> Mudballs supports or wishes to support ECL (http://
> ecls.sourceforge.net)

Indeed I do. I have unfortunately made a couple of decisions when
implementing
mudballs that reduced its portability but I'm sure we can work around
these
(or remove them in some cases).

I'm under the impression that they are limited to MOP usage and the
long form of
define-method-combination, the latter of which I am quite happy to
remove, but
there may be others lurking in corners. I'm busy building a more
recent version of ECL to see
where it breaks but I'm not too well versed in ECL error messages so any
incompatibilities that you can see would help.

sean.

[Juanjo]

On Mar 25, 12:45 pm, Sean Ross <ros...@gmail.com> wrote:
> I'm under the impression that they are limited to MOP usage and the  
> long form of define-method-combination

The latter should definitely work, and regarding MOP I think we have
almost everything except for things related to make-lambda and parts
that control the actual code in methods. In either case I will be glad
to fix or complete the missing pieces.

> there may be others lurking in corners. I'm busy building a more  
> recent version of ECL to see
> where it breaks but I'm not too well versed in ECL error messages so any
> incompatibilities that you can see would help.

Beware that I am in the process of merging a ton of changes in the git
and CVS trees. I want to produce a new release by the beginning of
next week. Seems that the linux and OS X ports are working right now,
though.

Juanjo