[Scilab-Dev] PVM

[Sylvestre Ledru]

> And what about to put the pvm sources into the scilab tree and compile 
> this sources if configure can't find the pvm .so library ?
> Like in scilab-4.1.2 ?
No way :p
I explain you why.

First, if you start to embed third party sources into your source tree,
where will it end ? 
You know our dependency list [1]; why should I put PVM and not FFTW,
matio or PCRE ?
(btw, I am too unhappy about having lapack & blas into the source tree
to add more dep into Scilab)

Second, when you incorporate third party code, you are more entitle to
edit the sources when you find a bug, instead of submitting a bug report
and waiting for the new upstream version (and consequently having the
risk of divergence.

Third, you have to integrate the source into your compilation process.
Autotools being everything but standard and straightforward, you have to
redo what they have done.

Have I been convincing enough ? :)


> By the way, PVM is working really fine with the patch 
The patch from JPC ?
I have to apply it in the trunk.

> and I am about to 
> have a parallel genetic algorithm (thanks to a student).
Cool, I am interested to see the result !

Sylvestre
[1] http://wiki.scilab.org/Dependencies_of_Scilab_5.X