How can one recompile a standard package (without rebuilding the whole thing) ?

[Emmanuel Charpentier]

Case in point : Trac#19469. This ticket will solve an annoying quirk un the Ipython notebook.

I tested it successfully on one installation, by recompiling the whole hog ( make distclean && make ).

I have another installation, now up to 6.10beta1. I did successfully :
git trac checkout
make (rebuilds the documentation, by the way...)

The resulting Sage still has the "old" behaviour.

I also tried ./sage -b in $SAGE_ROOT. Same result.

Is there a way to force the recompilation-reinstallation of the Ipythobn notebook ? I know that neither ./sage -i notebook or ./sage -f notebook work : these calls fail with complaining that they cannot find the source tarball... Further attempts to use this installation fail with the same error (sage has somehow registered that the Ipytho notebook is missing and tries to install it, unsuccessfully...).

What am I missing ?

[John H Palmieri]

On Wednesday, October 28, 2015 at 10:19:51 AM UTC-7, Emmanuel Charpentier wrote:

Case in point : Trac#19469. This ticket will solve an annoying quirk un the Ipython notebook.

I tested it successfully on one installation, by recompiling the whole hog ( make distclean && make ).

I have another installation, now up to 6.10beta1. I did successfully :
git trac checkout
make (rebuilds the documentation, by the way...)

Try

./configure
make

(although there is an open ticket which would make ./configure unnecessary).

  John

[Jeroen Demeyer]

On 2015-10-28 18:32, John H Palmieri wrote:
> (although there is an open ticket which would make ./configure unnecessary).

That's #19443, which needs review.

[Emmanuel Charpentier]

Mirabile dictu, it worked ! But it's a strange thing : AFAIK, ./configure is used to adapt a source to the specifics of a platform. This (usually) does not change between recompilations.

Except that Sagemath source carries a lot of its own buildig platform (e. g. gcc...). Is that the explanation ?

Thanks, John !

[Emmanuel Charpentier]

Thanks, Jeroen ! I'll have a look at this : I still have yet another machine to upgrade to #19469 : I'll try to install #19443 first, and use this to update #19469 as a test.

HTH,

--
Emmanuel Charpentier

[John H Palmieri]

On Wednesday, October 28, 2015 at 11:48:13 AM UTC-7, Emmanuel Charpentier wrote:

Mirabile dictu, it worked ! But it's a strange thing : AFAIK, ./configure is used to adapt a source to the specifics of a platform. This (usually) does not change between recompilations.

Except that Sagemath source carries a lot of its own buildig platform (e. g. gcc...). Is that the explanation ?

 
I believe that the configure script detects the version of each Sage package it needs to install, so if you want to update a package, you need to run ./configure so that it knows to build the new one instead of the old one.

  John

 

[Emmanuel Charpentier]

Done. 6.1beta2 + Trac#19443 + Trac#19469 passes ptestlong ==> positive review for Trac#19469.

HTH,

--
Emmanuel Charpentier