[17:36] <mgz> hm, looking at this osutils path manip stuff makes me want to resurrect the old Just van Rossum plan for Path objects.
[17:37] <mgz> if we have to make everyone use the osutils versions already we could at least provide an api that doesn't suck quite so much.
[22:30] <poolie> hi all
[22:44] <peitschie> hi poolie
[22:49] <elmo> is it relatively easy to get the modification time of files that have changed (i.e. would appear in bzr st output) through bzrlib or should I just parse the output of bzr st?
[22:53] <mwhudson> elmo: getting the mtime from a path is pretty easy
[22:53] <mwhudson> i think getting the list of files that would appear in st might be a little annoying
[22:55] <spiv> elmo: the_tree.get_file_mtime(file_id)
[22:56] <poolie> hi peitschie, elmo, spiv, mwhudson
[23:47] <spiv> poolie: if you can get the code out of the error class without regression performance then that would be nice
[23:47] <poolie> i'd much rather regress the feature than the performance
[23:47] <poolie> after all this isn't even something someone specifically asked for
[23:47] <spiv> That's true.
[23:48] <poolie> it was kind of inferred from "they wouldn't have got confused if ..."
[23:49] <spiv> I think maybe it's occasionally come up in user support on IRC too, but only very occasionally.
[23:49] <poolie> oh, seeing that reminded me of the url escaping thing too
[23:51] <spiv> I'm personally pretty relaxed about it now that the open_repository call is confined to str(e).  So constructing and repr()ing the exception is simple and safe, and it's only when generating text for display to a user we do the extra work (although I can see an argument for making that be something other than str(e) to be more careful).
[23:51] <spiv> But if you'd rather remove this marginal feature, that's fine with me too.
[23:53] <poolie> i guess it's dropped down my priority list too now
[23:53] <poolie> i would really rather not see code like that come in in future though
[23:53] <poolie> i think exceptions really should be value objects, or very close to it
[23:53] <poolie> (i realize it's a tradeoff)
[23:55] <spiv> Perhaps if we had other exceptions that wanted to provide that sort of hint when displayed to a user we'd find a better a home for the hint-generation/error-formatting.