Repository navigation
rework configuration system to not dangerously save state #1555
Description
The specific plans for this are detailed in an astropy-dev post: https://groups.google.com/forum/#!topic/astropy-dev/KW1bf4bLGW4
Activity
embray commentedon Oct 11, 2013
I'm not sure I'd necessarily title this specifically "to not use .cfg file" since I don't think that's settled. I believe there are cases for a config file, just that there are certain settings--specifically those that affect science results--that don't belong there. Maybe just "rework configuration system" period.
eteq commentedon Oct 15, 2013
Well, in my view the most important aspect is to not save content necessary for reproducing science. So I renamed it somewhat more specifically.
eteq commentedon Oct 15, 2013
And just to be clear, the particular solution of removing
astropy.cfgalso comes with some other advantages in the setup stageastrofrog commentedon Dec 1, 2013
@eteq @mdboom - would it make sense to draft an APE for this? I think there were some loose ends in the astropy-dev discussion, so maybe we can start to formalize the plan of action? We should get this done fairly soon in my opinion, so that it gets tested for a few months before the next release.
mdboom commentedon Dec 2, 2013
Yes -- I agree it should have an APE. I may be able to write a first draft of this later in the week, but not much before then.
eteq commentedon Dec 3, 2013
Agree that an APE is a good idea - "later this week" is perfectly fine as far as I'm concerned.
astrofrog commentedon May 6, 2014
@mdboom - closed by #2094?
embray commentedon May 6, 2014
I would certainly say so.