Thank you Dima ! I'll try that.

--
Emmanuel Charpentier

Le mercredi 28 septembre 2016 23:24:40 UTC+2, Dima Pasechnik a écrit :
>
> if you want to use system's BLAS/ATLAS, set SAGE_ATLAS_LIB to point to the 
> appropriate location
> (usually, /usr/lib, I suppose), and rebuild from scratch
> for more details see 
> http://doc.sagemath.org/html/en/installation/source.html#environment-variables
>
> It is a bug in the R package in question that it has hardcoded the 
> location of BLAS to the system's one.
>
>
> On Wednesday, September 28, 2016 at 9:14:00 AM UTC, Emmanuel Charpentier 
> wrote:
>>
>> A nice little puzzle for Sage's R users trying to use R 3.3.1 
>> (forthcoming in Trac#20523 <https://trac.sagemath.org/ticket/20523>) : 
>> the (still damnably useful, notwithstanding rstan) rjags package no 
>> longer installs.
>>
>> rjags calls the system's jags binary and system's jags modules (compiled 
>> as .so objects in Unix). However, there is currently (i. e. sage 7.4 
>> beta4) a conflict between the system's and Sage's notion of what BLAS 
>> should do : my (Debian) system's BLAS wants to use  a gotoblas function, 
>> not found in what Sage uses for BLAS.
>>
>> A possible workaround is to install a Sage-specific version of JAGS from 
>> the source, configured with something like "./configure 
>> --prefix=$SAGE_ROOT/local") in a Sage shell (and compiled and installed 
>> in the Sage shell...). This worked like a charm for me.
>>
>> I do not know Sage's build system well enough to know if forcing Sage to 
>> use system's numerical livraries (by forcing, e. g. to use the local ATLAS 
>> libraries) should be enough.
>>
>> HTH,
>>
>> --
>> Emmanuel Charpentier
>>
>>

-- 
You received this message because you are subscribed to the Google Groups 
"sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/sage-support.
For more options, visit https://groups.google.com/d/optout.

Reply via email to