16:47:55 <Heffalump> vifon: http://bugs.darcs.net/issue2657
16:49:44 <vifon> Heffalump: I think it's also worth mentioning the transitive dependencies are also shown on the graph as direct ones in 2.16, which is much worse than crossing the tag boundary.
16:50:09 <Heffalump> vifon: fair point. I should write another test script :-)
16:50:20 <Heffalump> it's a bit harder to spot with a trivial grep
16:50:38 <vifon> E.g. A>B>C would also show a A>C dependency on 2.16 but not on 2.14.
16:51:41 <vifon> BTW, can I force Darcs (2.14) to show the whole graph regardless of tags? --last 9999 works but feels hacky.
16:53:08 <Heffalump> I can't think of a better way, from reading the docs. I guess that issue validates Ben's point about removing the "default up to the last tag" feature.
16:53:58 <vifon> Yeah. Though I do not agree with him on the obliterate example.
16:54:45 <vifon> Since a clean tag depends on everything after it, you need to obliterate the tag too. And that not something you'd do lightly.
16:55:14 <Heffalump> the change would certainly make obliterate -a useless
16:55:14 <vifon> At least from my very short experience, this limitation does make sense. And is probably a vast optimization.
16:55:30 <Heffalump> at the moment I actually do use it sometimes to jump back a long way but not forever
16:56:21 <Heffalump> yeah - it is indeed a significant optimisation, but it leaks implementation details into the UI somewhat
16:56:42 <Heffalump> it would be nice to find a more user-friendly approach/heuristic
16:57:06 <vifon> I'd consider a tag a closed chapter. You work on everything that happened since the tag, and only on it.
16:57:25 <vifon> In my opinion this behaviour is fine.
16:57:30 <Heffalump> except that they aren't totally ordered, and it's not normally visible to you whether they are clean or not
16:57:41 <vifon> True.