[gentoo-science] [sage-on-gentoo] sage-clib(mpir)

[gentoo-science] [sage-on-gentoo] sage-clib(mpir)

[Steven Trogdon]

[-- Attachment #1: Type: text/plain, Size: 604 bytes --]

Hey Guys,

Perhaps I shouldn't have sage-clib with use=mpir enabled, but if I do  
the
sage-clib-4.4.4-importenv.patch fails because the first hunk in
sage-clib-4.4.3-replace-gmp-with-mpir.patch has already overwritten gmp  
with mpir. I'm not sure
but perhaps the sage-clib-4.4.3-replace-gmp-with-mpir.patch should also  
have the environment stuff that's in the first hunk of  
sage-clib-4.4.4-importenv.patch.

Steve

P.S.
The parallel build of sage-core now only takes several minutes here  
with 4 cpu cores. Of course ccache was enabled so a fresh build would  
take a bit longer.

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-science] [sage-on-gentoo] sage-clib(mpir)

[François Bissey]

> Hey Guys,
> 
> Perhaps I shouldn't have sage-clib with use=mpir enabled, but if I do
> the
> sage-clib-4.4.4-importenv.patch fails because the first hunk in
> sage-clib-4.4.3-replace-gmp-with-mpir.patch has already overwritten gmp
> with mpir. I'm not sure
> but perhaps the sage-clib-4.4.3-replace-gmp-with-mpir.patch should also
> have the environment stuff that's in the first hunk of
> sage-clib-4.4.4-importenv.patch.
> 
Hi Steve,

we probably should disable that flag until we have a number of kinks sorted
out....
Patches and dependencies - in theory we would need to enforce pari[mpir=]
and ntl[mpir=]. I will look at these patches later, after I have gone through
my other priorities:
*ATLAS independence [hoping it will hit the overlay in a few hours]
*system singular which has to be read for sage-4.5
 
> 
> P.S.
> The parallel build of sage-core now only takes several minutes here
> with 4 cpu cores. Of course ccache was enabled so a fresh build would
> take a bit longer.
Yeah! With real CPUs that must be awesome. The cython calls used to be 
serial and use only one CPU, ccache doesn't do anything about those, so
you should have a real benefit on that end. Then the compilation will be
parallelized which is good as well.

Francois