[05:37] <thomi> lifeless: got any advice for how to responsibly deprecate parts of a python module over time? Any hints / pointers?
[05:38] <thomi> I'm sure I've seen people use a '@deprecated' decorator, not sure how effective that is though
[05:56] <lifeless> thomi: rm ? :)
[05:56] <thomi> heh
[05:57] <thomi> I guess I was looking for a nice way to phase out old functions that I don't want people using.
[05:57] <thomi> Are you suggesting that it's better to just bite the bullet and delete them and deal with the fallout?
[06:00] <lifeless> Well
[06:00] <lifeless> So there are a few nuances
[06:00] <lifeless> do you have an ecosystem?
[06:00] <lifeless> How much time do folk in that ecosystem have? How much will it annoy them having to deal with deprecations?
[06:00] <lifeless> For instance, if you have a three tier thing
[06:00] <lifeless> core -> plugins -> users
[06:01] <lifeless> then when core deprecates something, and then removes, plugins have a grace period, which can be good
[06:01] <lifeless> but if you have
[06:01] <lifeless> core -> users
[06:01] <lifeless> then deprecation doesn't really do anything
[06:01] <lifeless> as the folk affected are the folk with the power and expectation of fixing.
[06:02] <lifeless> bzr has some decorators you could snarf
[19:48] <mwhudson> morning
[19:57] <chilts> morning
[20:03] <ojwb> morning
[20:16] <thomi> lifeless: sorry, I had to dissapear yesterday - thanks for your answer
[20:16] <thomi> I'm talking about autopilot, so I have core -> users
[20:20] <Atamira> moring
[20:20] <Atamira> morning
[20:25] <ajmitch> morning
[21:04] <fmarier> has everybody seen this already? http://no.softwarepatents.org.nz/
[21:04] <lifeless> I sure hope so !
[21:04] <fmarier> lifeless: me too, but just in case :)
[21:04] <ajmitch> s/seen/signed/
[21:04] <lifeless> thomi: so yeah, changing an API on your users is much less traumatic when they aren't producing other components in the ecosystem
[21:05] <thomi> lifeless: sounds good to me. Especially since (so far), I'm a developer on all the projects that use autopilot
[21:06] <ajmitch> that makes it a bit easier to coordinate deprecations
[21:06] <thomi> yeah
[21:06] <lifeless> for direct use in that situation I wouldn't deprecate
[21:07] <lifeless> deprecation in e.g. bzrlib is much more important IMNSHO.
[21:07] <lifeless> of course, if you want to grow an ecosystem - chicken and egg :)
[21:08] <thomi> yeah