Rationalize the number of ancillary (default matplotlibrc) files #1123

Closed
pelson opened this Issue Aug 21, 2012 · 3 comments

Projects

None yet

3 participants

@pelson
Member
pelson commented Aug 21, 2012

#1114 highlighted some redundancy with regards to the contents of the default matplotlibrc file. There are at least 3 copies of a "default" matplotlibrc:

matplotlibrc.template
./lib/matplotlib/mpl-data/matplotlibrc
./doc/pyplots/matplotlibrc

Replace these with a single definitive source.

Additionally, investigate the use of lib/matplotlib/mpl-data/matplotlib.conf to confirm that it is still needed.

@efiring
Member
efiring commented Aug 21, 2012

lib/matplotlib/mpl-data/matplotlib.conf and matplotlibrc are not revision-controlled (they are listed in .gitignore) but are built from templates whenever setup.py is run.
matplotlib.conf and its template look horribly obsolete; my guess is that they were part of an experiment with using an ini-style configuration, and were forgotten when the experiment was abandoned. So, I think that setup.py needs to be modified to no longer build matplotlib.conf, and then matplotlib.conf.template can be deleted from the repo.

The building of matplotlibrc in mpl-data from matplotlibrc.template is essential, and matplotlibrc.template is what needs to be modified when a new rc param is added.

doc/pyplots/matplotlibrc is there to set the backend to Agg and the figsize to (6,4) for generating documentation plots. Everything else is commented out, and some of that is obsolete. This fill should be trimmed down to include only the two settings actually being used.

All this would be a nice cleanup for 1.2. If there is agreement that I have described the situation correctly, I can make a pull request for it.

@pelson
Member
pelson commented Aug 21, 2012

@efiring: Thank you. I forgot to check what was built and what was revisioned.

There is a plot directive option which can be put in conf.py to handle the figure size if I remember correctly. Can that be used instead? (I'm not sure about defining the backend though).

I would go ahead and make the proposed changes and submit as PR anyway. If anyone has an exception to your statements, it can be made against the appropriate changes. I've put this issue into the 1.2.x milestone.

@mdboom
Member
mdboom commented Aug 21, 2012

I think @efiring's plan sounds right.

@efiring efiring closed this in 477afb7 Aug 21, 2012
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment