matplotlib 1.0.1rc available for testing, building

We are long overdue on getting a bugfix release of 1.0.0 out, so I have uploaded an rc for testing at

https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.0.1/

Christoph and Russell – if you have time could you build win32 and OSX binaries for testing as well. I don’t believe either of you have developer permissions to upload directly to this site, but I would be happy to add you if you send me an sf id. Alternatively, you can upload them to a site of your choosing and I’ll upload them for you (drop.io was acquired by facebook and no longer works).

There are a number of bugs and patches on the tracker that it would have been nice to tackle before this release, but there have been enough improvements in the branch that delaying at this point would be a case of the perfect being the enemy of the good so I think we should move forward now. Nonetheless, if there are important bugs or patches that anyone can tackle before we cut the final release, that would be great.

In sf, we used to be able to tag a file as the preferred file for a given OS, but in the new file manager interface I no longer see a way to do this, so that for example the rc files don’t show up as the default download options. Does anyone know how to do this, or perhaps someone can suggest a drop.io replacement that supports multiple user uploaders (eg the different binary builders) and public, no-registration downloads. I googled a bit and didn’t find anything that fit the bill.

JDH

Hi John,

···

On Sun, Jan 2, 2011 at 18:52, John Hunter <jdh2358@...149...> wrote:

We are long overdue on getting a bugfix release of 1.0.0 out, so I have
uploaded an rc for testing at

matplotlib - Browse /matplotlib/matplotlib-1.0.1 at SourceForge.net

Maybe I just missed them, but I can't find in the tarball the data
files needed to run the examples without internet connection. I
thought it was decided to ship them directly in the tarball and so use
the examples.directory rc option to point to them at build-time and
avoid to download, but if they're not in the tarball.... :slight_smile:

Cheers,
--
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi

I just added a sample data release file to the same dir – mpl_sampledata-1.0.1rc.tar.gz - different tarball, but should contain everything you need. Will this meet Debian’s requirements? We could add it to the main tarball, but part of the original impetus in moving the data out of the main tree was to make the main tarball smaller. I can go either way if it is a major hassle for you guys to do it differently.

JDH

···

On Sun, Jan 2, 2011 at 4:25 PM, Sandro Tosi <morph@…12…> wrote:

https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.0.1/

Maybe I just missed them, but I can’t find in the tarball the data

files needed to run the examples without internet connection. I

thought it was decided to ship them directly in the tarball and so use

the examples.directory rc option to point to them at build-time and

avoid to download, but if they’re not in the tarball… :slight_smile:

Thanks for the fast reply! There is a quite-new feature that allows
use to use several tarballs to create a single source package (in this
case the source & sample_data tarballs to generate the matplotlib
debian source package): I can try that, and I'll let you know if I
fail or not.

Cheers,

···

On Sun, Jan 2, 2011 at 23:50, John Hunter <jdh2358@...149...> wrote:

I just added a sample data release file to the same dir --
mpl_sampledata-1.0.1rc.tar.gz - different tarball, but should contain
everything you need. Will this meet Debian's requirements? We could add it
to the main tarball, but part of the original impetus in moving the data out
of the main tree was to make the main tarball smaller. I can go either way
if it is a major hassle for you guys to do it differently.

--
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi

You might consider looking at my Ubuntu PPA[1] if you are going to work on
packaging. I seem to have everything working at this point, so perhaps
that can save you some work.

Cheers,

- Ben

[1] OpenID transaction in progress

···

On Sun, 2 Jan 2011 23:25:12 +0100, Sandro Tosi <morph@...12...> wrote:

Maybe I just missed them, but I can't find in the tarball the data
files needed to run the examples without internet connection. I
thought it was decided to ship them directly in the tarball and so use
the examples.directory rc option to point to them at build-time and
avoid to download, but if they're not in the tarball.... :slight_smile:

I already looked at it, but it still needs some work.

Regards,

···

On Mon, Jan 3, 2011 at 00:21, Ben Gamari <bgamari.foss@...149...> wrote:

On Sun, 2 Jan 2011 23:25:12 +0100, Sandro Tosi <morph@...12...> wrote:

Maybe I just missed them, but I can't find in the tarball the data
files needed to run the examples without internet connection. I
thought it was decided to ship them directly in the tarball and so use
the examples.directory rc option to point to them at build-time and
avoid to download, but if they're not in the tarball.... :slight_smile:

You might consider looking at my Ubuntu PPA[1] if you are going to work on
packaging. I seem to have everything working at this point, so perhaps
that can save you some work.

--
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi

I found the answer to this – I don’t know why it wasn’t working for me yesterday. In the file manager interface there is a little black icon with a circle around “i”, presumably for “info”) in which you can check a button indicating that this is the default download for a given platform. I could swear I tried this yesterday…

JDH

···

On Sun, Jan 2, 2011 at 11:52 AM, John Hunter <jdh2358@…552…149…> wrote:

In sf, we used to be able to tag a file as the preferred file for a given OS, but in the new file manager interface I no longer see a way to do this, so that for example the rc files don’t show up as the default download options.

I just build and uploaded a version for python 2.6. I have tested it on Intel 10.4 and plan to test it on 10.5 and 10.6 before trying to build a version for 32-bit Python 2.7. I will keep you posted.

Regards,

– Russell

···

On Jan 2, 2011, at 9:52 AM, John Hunter wrote:

We are long overdue on getting a bugfix release of 1.0.0 out, so I have uploaded an rc for testing at

https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.0.1/

Christoph and Russell – if you have time could you build win32 and OSX binaries for testing as well. I don’t believe either of you have developer permissions to upload directly to this site, but I would be happy to add you if you send me an sf id. Alternatively, you can upload them to a site of your choosing and I’ll upload them for you (drop.io was acquired by facebook and no longer works).

There are a number of bugs and patches on the tracker that it would have been nice to tackle before this release, but there have been enough improvements in the branch that delaying at this point would be a case of the perfect being the enemy of the good so I think we should move forward now. Nonetheless, if there are important bugs or patches that anyone can tackle before we cut the final release, that would be great.

In sf, we used to be able to tag a file as the preferred file for a given OS, but in the new file manager interface I no longer see a way to do this, so that for example the rc files don’t show up as the default download options. Does anyone know how to do this, or perhaps someone can suggest a drop.io replacement that supports multiple user uploaders (eg the different binary builders) and public, no-registration downloads. I googled a bit and didn’t find anything that fit the bill.

JDH

The news:

  • The Mac binary installer for python.org Python 2.6 works on my Mac OS X 10.4 and 10.5 machines (both Intel). However, it segfaults on my 10.6 machine if there is no existing ~/.matplotlib and ~/.fontconfig. If others could test this it would be most helpful. I have not yet tried it on 10.3.9.

To reiterate: a proper test is:

  • delete ~/.matplotlib if it exists

  • delete ~/.fontconfig if it exists

  • run python and try to import pylab or matplotlib

– Russell

···

On Jan 2, 2011, at 9:52 AM, John Hunter wrote:

We are long overdue on getting a bugfix release of 1.0.0 out, so I have uploaded an rc for testing at

https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.0.1/

Christoph and Russell – if you have time could you build win32 and OSX binaries for testing as well. I don’t believe either of you have developer permissions to upload directly to this site, but I would be happy to add you if you send me an sf id. Alternatively, you can upload them to a site of your choosing and I’ll upload them for you (drop.io was acquired by facebook and no longer works).

There are a number of bugs and patches on the tracker that it would have been nice to tackle before this release, but there have been enough improvements in the branch that delaying at this point would be a case of the perfect being the enemy of the good so I think we should move forward now. Nonetheless, if there are important bugs or patches that anyone can tackle before we cut the final release, that would be great.

In sf, we used to be able to tag a file as the preferred file for a given OS, but in the new file manager interface I no longer see a way to do this, so that for example the rc files don’t show up as the default download options. Does anyone know how to do this, or perhaps someone can suggest a drop.io replacement that supports multiple user uploaders (eg the different binary builders) and public, no-registration downloads. I googled a bit and didn’t find anything that fit the bill.

JDH

matplotlib 1.0.1rc has pytz 2010h but the current version is 2010o. (dateutil is current at 1.5).

Should I report this as a bug or is this email sufficient?

-- Russell

email is sufficient. I’m inclined to fix this for the trunk (targeting a 1.1 release) rather than the branch as we are aiming for maximum stability in the branch and 1.0.1 bugfix release. However, if there are known bugs in the wild vis-a-vid mpl that the old version of pytz is causing us, I am happy to reconsider.

JDH

···

On Mon, Jan 3, 2011 at 2:00 PM, Russell Owen <rowen@…907…> wrote:

matplotlib 1.0.1rc has pytz 2010h but the current version is 2010o.

(dateutil is current at 1.5).

Should I report this as a bug or is this email sufficient?

These packages should be kept up-to-date if possible. There is an old bug report that was closed.

<http://sourceforge.net/tracker/index.php?func=detail&aid=3009273&group_id=80706&atid=560720&gt;

···

On 1/3/2011 12:00 PM, Russell Owen wrote:

matplotlib 1.0.1rc has pytz 2010h but the current version is 2010o.
(dateutil is current at 1.5).

Should I report this as a bug or is this email sufficient?

-- Russell

--
Christoph

If you run an example with --verbose-debug-annoying with a clean cache, you’ll get more information about what font file it is parsing right before the crash. Not sure if this will help, but it would be interesting to see.

JDH

···

On Mon, Jan 3, 2011 at 1:57 PM, Russell Owen <rowen@…907…> wrote:

The news:

  • The Mac binary installer for python.org Python 2.6 works on my Mac OS X 10.4 and 10.5 machines (both Intel). However, it segfaults on my 10.6 machine if there is no existing ~/.matplotlib and ~/.fontconfig. If others could test this it would be most helpful. I have not yet tried it on 10.3.9.

To reiterate: a proper test is:

  • delete ~/.matplotlib if it exists
  • delete ~/.fontconfig if it exists
  • run python and try to import pylab or matplotlib

– Russell

Since the version is out of date and may contain bad tz or tzinfo, I’m willing to consider it a bug and fixed it in the branch. Seems to be working fine on my tests.

···

On Mon, Jan 3, 2011 at 2:19 PM, Christoph Gohlke <cgohlke@…244…> wrote:

These packages should be kept up-to-date if possible. There is an old

bug report that was closed.

<http://sourceforge.net/tracker/index.php?func=detail&aid=3009273&group_id=80706&atid=560720>