MANIFEST.in

Charlie is fixing a few problems with the release, and noticed that
many of the new directories and files that have been created of late
were not added to MANIFEST.in, which is used by sdist in the build
process to build a correct source distribution. He added a few things
(eg ttconv and setup.cfg.template) but please take a look over this
file if you have made significant additions or reorganizations in the
last 6 months to make sure everything is there. It picks up pattern
matches recursively so it is usually enough to update with a new dir
or wildcard, but please maintain this file as you contribute. It's
easy to forget if you are not involved in the build, so I'll update
the CODING_GUIDE

JDH

I'm checking the matplotlib-0.91.1 gentoo ebuild that Sébastien Fabbro put
together. After installing, I get a number of errors when running
backend_driver.py that are probably related to MANIFEST.in: missing pyplot
and npyma. There are many other errors exposed by backend_driver as well, but
maybe some will clear up with an improved MANIFEST.

Charlie, will you have time to upload an improved src distribution any time
soon?

Darren

···

On Tuesday 04 December 2007 9:18:14 am John Hunter wrote:

Charlie is fixing a few problems with the release, and noticed that
many of the new directories and files that have been created of late
were not added to MANIFEST.in, which is used by sdist in the build
process to build a correct source distribution. He added a few things
(eg ttconv and setup.cfg.template) but please take a look over this
file if you have made significant additions or reorganizations in the
last 6 months to make sure everything is there. It picks up pattern
matches recursively so it is usually enough to update with a new dir
or wildcard, but please maintain this file as you contribute. It's
easy to forget if you are not involved in the build, so I'll update
the CODING_GUIDE

Darren Dale wrote:

I'm checking the matplotlib-0.91.1 gentoo ebuild that S�bastien Fabbro put together. After installing, I get a number of errors when running backend_driver.py that are probably related to MANIFEST.in: missing pyplot and npyma. There are many other errors exposed by backend_driver as well, but maybe some will clear up with an improved MANIFEST.

pyplot.py is in lib, and MANIFEST.in already recursively includes everything in lib, so I don't think this is the problem.

npyma is inside of lib/numerix/, which again should not have required any change to MANIFEST.in.

Eric

Could you post some of the errors?

Thanks,
JDH

···

On Dec 4, 2007 2:17 PM, Darren Dale <darren.dale@...143...> wrote:

I'm checking the matplotlib-0.91.1 gentoo ebuild that Sébastien Fabbro put
together. After installing, I get a number of errors when running
backend_driver.py that are probably related to MANIFEST.in: missing pyplot
and npyma. There are many other errors exposed by backend_driver as well, but
maybe some will clear up with an improved MANIFEST.

Damn thats frustrating. I just repeated the same steps I took before I wrote
about finding a problem, and now everything is ok. Sorry for the noise.

···

On Tuesday 04 December 2007 4:06:19 pm John Hunter wrote:

On Dec 4, 2007 2:17 PM, Darren Dale <darren.dale@...143...> wrote:
> I'm checking the matplotlib-0.91.1 gentoo ebuild that Sébastien Fabbro
> put together. After installing, I get a number of errors when running
> backend_driver.py that are probably related to MANIFEST.in: missing
> pyplot and npyma. There are many other errors exposed by backend_driver
> as well, but maybe some will clear up with an improved MANIFEST.

Try flushing ~/.matplotlib and retest -- some of the bugs may be
masked by the cache

JDH

···

On Dec 4, 2007 4:16 PM, Darren Dale <darren.dale@...143...> wrote:

Damn thats frustrating. I just repeated the same steps I took before I wrote
about finding a problem, and now everything is ok. Sorry for the noise.

Done, no errors.

···

On Tuesday 04 December 2007 5:21:38 pm John Hunter wrote:

On Dec 4, 2007 4:16 PM, Darren Dale <darren.dale@...143...> wrote:
> Damn thats frustrating. I just repeated the same steps I took before I
> wrote about finding a problem, and now everything is ok. Sorry for the
> noise.

Try flushing ~/.matplotlib and retest -- some of the bugs may be
masked by the cache