C Y wrote:
>--- "Viktor T. Toth" wrote:
>
>
>
>>Actually, I have another question in the same vein: Why is it
>>necessary for Maxima to be compatible with every LISP compiler out
>>there? Just to be clear, I don't mind it (indeed, using different
>>LISPs proved to be quite a good way to catch some subtle errors),
>>I'm just curious. Are there really people out there who can download
>>and install Maxima but not, say, GCL, or is there some other reason
>>why cross-compiler compatibility is necessary? (I have a feeling
>>that I'm missing something obvious here that should have hit
>>me right in the nose.)
>>
>>
>
>No, not really. The reason it's a project goal to compile on as many
>lisps as possible is it reduces our dependancy on any one of them, in
>addition to the testing advantages you already mentioned.
>
I think that this goal has been a substantial distraction, but as long
as it hasn't required me
to do much, I can't really argue. I think that having a downloadable
binary on one
open-source Lisp is really not too much of a dependency. For my own
work I either
use such a download or I compile/run Maxima in Allegro Common Lisp, which
is NOT free, but has some advantages. (And there is a site license at
my school).
> So if a lisp
>project should fade away for some reason (not that gcl, clisp, cmucl,
>sbcl, etc show any indications of that) we would not be hurting trying
>to make it work elsewhere.
>
This seems quite remote.
> On Windows, GCL is still the best
>performing default choice with Clisp being the only free alternative
>(at least until the sbcl port arrives.)
>
I think the sbcl port is a distraction, but again, if the people doing
that are not hurting
other efforts, and they weren't going to be doing anything specific to
improve Maxima,
I can't object.
> There are some occasional
>performance reasons as well, and some custom things that can sometimes
>be done in one lisp but not others (for example, anyone wanting to
>explore an McCLIM interface for Maxima would have to do a lot of work
>to port it to GCL if that was our only lisp).
>
>
>So there is no immediate reason, aside from testing and maybe certain
>performance critical situations, but it provides a "warm and fuzzy"
>feeling knowing we have many alternatives to work with.
>
>
One downside to all the ports is that anyone reading this news list will
get the
feeling that Maxima is in its infancy because of so many messages
talking about
problems porting it to lin*x, GCL version x.y.z, etc.
>CY