Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread François Bissey
On 11/10/15 08:09, John H Palmieri wrote: With the current version, running grep -R 'width and height must' local/lib/python/site-packages/matplotlib* finds a match. What happens if you do the same with the new version? Blank: fbissey@QCD-nzi3 ~/sandbox/git-fork/sage $ grep -R 'width and

Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread François Bissey
On 11/10/15 08:12, John H Palmieri wrote: Oh, and in original tarball, it's in 'src/_backend_agg.cpp'. I'll have to add that between 1.4.3 and 1.5.0 'src/_backend_agg.cpp' has completely been rewritten... fbissey@QCD-nzi3 ~ $ wc

Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread John H Palmieri
With the current version, running grep -R 'width and height must' local/lib/python/site-packages/matplotlib* finds a match. What happens if you do the same with the new version? On Monday, November 9, 2015 at 10:22:12 AM UTC-8, François wrote: > > I tried a few already but I can try some

Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread Francois Bissey
I tried a few already but I can try some more. > On 10/11/2015, at 04:39, kcrisman wrote: > > > >> That test now just pass and I have no idea what the new limit is if > >> any. I'll note that it followed by a number of examples that are not > >> tested as they lead to

Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread kcrisman
> > > >> That test now just pass and I have no idea what the new limit is if > >> any. I'll note that it followed by a number of examples that are not > >> tested as they lead to segfault. After the upgrade they are still > >> segfaulting. Should I just remove the above doctest as it is not >

Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread John H Palmieri
Oh, and in original tarball, it's in 'src/_backend_agg.cpp'. John On Monday, November 9, 2015 at 11:09:49 AM UTC-8, John H Palmieri wrote: > > With the current version, running > > grep -R 'width and height must' > local/lib/python/site-packages/matplotlib* > > finds a match. What happens

Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread John H Palmieri
The only possibly relevant thing I could find was in src/_image.h: $ grep 32768 _image.h if (rows >= 32768 || cols >= 32768) { throw "rows and cols must both be less than 32768"; But something like this was also in the old version... I also looked for "less than" and "below" in

Re: [sage-devel] upgrading to matplotlib

2015-11-09 Thread François Bissey
On 11/10/15 11:33, John H Palmieri wrote: The only possibly relevant thing I could find was in src/_image.h: $ grep 32768 _image.h if (rows >= 32768 || cols >= 32768) { throw "rows and cols must both be less than 32768"; But something like this was also in the old version... I

Re: [sage-devel] upgrading to matplotlib

2015-11-08 Thread Francois Bissey
> On 9/11/2015, at 16:03, Jeroen Demeyer wrote: > > On 2015-11-08 15:01, François Bissey wrote: >> That test now just pass and I have no idea what the new limit is if >> any. I'll note that it followed by a number of examples that are not >> tested as they lead to

[sage-devel] upgrading to matplotlib

2015-11-08 Thread François Bissey
Hi all, I am working on a series of ticket to upgrade numpy/matplotlib/scipy. http://trac.sagemath.org/ticket/17642 (numpy) http://trac.sagemath.org/ticket/19556 (matplotlib) http://trac.sagemath.org/ticket/17643 (scipy) The upgrade is developed as being a simultaneous one. I have a pending

Re: [sage-devel] upgrading to matplotlib

2015-11-08 Thread Jeroen Demeyer
On 2015-11-08 15:01, François Bissey wrote: That test now just pass and I have no idea what the new limit is if any. I'll note that it followed by a number of examples that are not tested as they lead to segfault. After the upgrade they are still segfaulting. Should I just remove the above