I have opened a Pull request (
https://github.com/matplotlib/matplotlib/pull/4305) to improve the error
message and avoid the comparison between a string and version number.
ons. 1. apr. 2015 kl. 18.05 skrev Christian Ambros :
> Hi,
>
> I installed the libfreetype6-dev package and than re-start
There were multiple PRs that toched the freetype handling (iirc from Jens
and Tony Yu), but I don't remember if either got mreged off the top of my
head.
On Wed, Apr 1, 2015 at 12:55 PM Benjamin Root wrote:
> Didn't we fix that in trunk recently?
>
> On Wed, Apr 1, 2015 at 12:51 PM, Jens Nielsen
Hi,
I installed the libfreetype6-dev package and than re-started the upgrade
process with sudo pip3 install matplotlib --upgrade, which took a while but
finally was successful.It seems to be all right now.
Thanks for the hint with freetype. That helped a lot.
cheers,
Christian --
"A little le
Didn't we fix that in trunk recently?
On Wed, Apr 1, 2015 at 12:51 PM, Jens Nielsen
wrote:
> Looking more closely at this I think it is a bug on our side. When
> freetype is not found it returns version as 'Failed to identify version.'
> which
> it tries to compare to a version number. The vers
Yeah, that mirrors what others have stated. The common thread seems to be
that all of these users were comfortable with doing "sudo pip install
" (myself included). I was in a rush when I originally encountered
issues back in the summer on my 12.04 machine, so I just switched to
miniconda and didn'
Looking more closely at this I think it is a bug on our side. When freetype
is not found it returns version as 'Failed to identify version.' which it
tries to compare to a version number. The version number is correctly
converted from string to numbers using loosversion but this string just
makes i
Hi,
as you can see: 14.3.1 which is the latest, because before I started upgrading,
I read about possible issues here and upgraded the setuptools as conclusion.
Python 3.4.0 (default, Apr 11 2014, 13:05:11)
[GCC 4.8.2] on linux
Type "help", "copyright", "credits" or "license" for more information
Good point. At least put up a decent message to help users understand what
is wrong.
On Wed, Apr 1, 2015 at 9:57 AM, Thomas Caswell wrote:
> Independent of what exactly is going wrong, the issue is that it _isn't_
> finding the right version of freetype and iirc LooseVersion is handling the
> 'n
Independent of what exactly is going wrong, the issue is that it _isn't_
finding the right version of freetype and iirc LooseVersion is handling the
'not found' return code badly. This is coming up often enough we probably
do need to special case this check with a try/except.
On Wed, Apr 1, 2015
Actually, look at the traceback... it is using distutils' version.py.
That's weird. Is that a result of setuptools monkey-patching?
On Wed, Apr 1, 2015 at 9:25 AM, Jens Nielsen wrote:
> I think we have seen this issue before and it seems to be caused by an out
> of date version of setuptools. I
I think we have seen this issue before and it seems to be caused by an out
of date version of setuptools. I tried reproducing it on fresh ubuntu 14.04
machine but was not able to reproduce the issue. Do you know which version
of setuptools you are using?
Jens
ons. 1. apr. 2015 kl. 14.19 skrev Tho
Make sure you have `freetype-dev` installed at the system level.
Tom
On Wed, Apr 1, 2015 at 8:02 AM Christian Ambros wrote:
> Hi,
>
> I'm facing the same trouble with installing matplotlib 1.4.3 and 1.5.dev1.
> running
>
> python3 setup.py build
>
> in the unarchived directory gives this:
>
Hi,
I'm facing the same trouble with installing matplotlib 1.4.3 and 1.5.dev1.
running
python3 setup.py build
in the unarchived directory gives this:
Edit setup.cfg to change the build options
BUILDING MATPLOTLIB
"please don't do that"
Yes, I will make sure that any antagonizing I do in the future, it will be
completely clear that I am the one doing it. ;-)
On Wed, Mar 18, 2015 at 2:04 PM, Thomas Caswell wrote:
> Hmm, I can't read and miss-attributed who was antagonizing Sandro, please
> don't do that.
Hmm, I can't read and miss-attributed who was antagonizing Sandro, please
don't do that.
On Wed, Mar 18, 2015 at 2:01 PM Benjamin Root wrote:
> For my part, I didn't take Keith's comment as antagonizing. If anything, I
> should apologize to Sandro. It was not necessary for me to drag Debian into
For my part, I didn't take Keith's comment as antagonizing. If anything, I
should apologize to Sandro. It was not necessary for me to drag Debian into
this, because all I know is that I was having issues on Ubuntu.
Ben Root
On Wed, Mar 18, 2015 at 1:51 PM, Thomas Caswell wrote:
> We do support
We do support ubuntu, travis.ci (which we use for continuous integration
testing) is ubuntu based and my main development box is ubuntu (but I
mostly work inside conda environments rather than virtualenvs these days).
Even though it is the worst thing for a dev to say, 'it works on my
machine'.
Pa
We would too. This is the first time I have seen updating setuptools not
work. That was the fix... I have no clue why it is broken on your system.
On Wed, Mar 18, 2015 at 1:20 PM, wrote:
> Ok, I will check out anaconda anyway.
>
>
>
> I would think that the matplotlib maintainers would want to m
One thing I just noticed is that python3.4 and the distutils libraries are
installed at /usr/lib/python3.4/distutils/, but the setuptools is located
at /usr/local/lib/python3.4/dist-packages/. One of the oddities of
setuptools is that it monkey-patches distutils, if I understand it
correctly, so pe
19 matches
Mail list logo
