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
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 I'm checking the matplotlib-0.91.1 gentoo ebuild that S=E9bastien Fabbro pu= t=20 together. After installing, I get a number of errors when running=20 backend_driver.py that are probably related to MANIFEST.in: missing pyplot= =20 and npyma. There are many other errors exposed by backend_driver as well, b= ut=20 maybe some will clear up with an improved MANIFEST. Charlie, will you have time to upload an improved src distribution any time= =20 soon? Darren
Darren Dale 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. 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
On Dec 4, 2007 2:17 PM, Darren Dale <dar...@co...> wrote: > I'm checking the matplotlib-0.91.1 gentoo ebuild that S=E9bastien Fabbro = put > together. After installing, I get a number of errors when running > backend_driver.py that are probably related to MANIFEST.in: missing pyplo= t > and npyma. There are many other errors exposed by backend_driver as well,= but > maybe some will clear up with an improved MANIFEST. Could you post some of the errors? Thanks, JDH
On Tuesday 04 December 2007 4:06:19 pm John Hunter wrote: > On Dec 4, 2007 2:17 PM, Darren Dale <dar...@co...> wrote: > > I'm checking the matplotlib-0.91.1 gentoo ebuild that S=E9bastien 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 wrot= e=20 about finding a problem, and now everything is ok. Sorry for the noise.
On Dec 4, 2007 4:16 PM, Darren Dale <dar...@co...> 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 JDH
On Tuesday 04 December 2007 5:21:38 pm John Hunter wrote: > On Dec 4, 2007 4:16 PM, Darren Dale <dar...@co...> 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 Done, no errors.