I'm not sure I would recommend this technique.) I did it because merge was giving me a result I didn't understand, but I did understand how it out to go, so export seemed to be a clean way of forcing the issue. I would imagine that as I understand svn better I won't do that any more. It's usually better to diagnose and do the right thing.
jim On Sep 20, 2006, at 9:54 AM, Benjamin Shine wrote: > In this case, the history for lpp-2587.lzx is not particularly > important -- it's a test file, after all, and the change history > was restructuring the test from the user-provided test, which is in > jira, to an automated test. > > But, as the guy who added lpp-2587.lzx, I am concerned about why it > got weird. Seems like only your very-carefulness rescued it. > > It feels like merging is more error-prone with svn than with > perforce. We knew this would be the case going into the change, > but, now that we're in it, ouch. > > I haven't seen export used in the context of a merge before. What > was the advantage of export over merge, like this: > svn merge -r1718:1719 http://svn.openlaszlo.org/openlaszlo/trunk/ > test/bugs/lpp-2587.lzx > > Could you update the wiki with what you've been learning lately > about merging? > > > On Sep 19, 2006, at 9:26 PM, Jim Grandy wrote: > >> Umm, glad you asked... as it turns out, I left two steps out of my >> narrative. The middle bit should have been: >> >> svn export -r 1765 http://svn.openlaszlo.org/openlaszlo/trunk/test/ >> bugs/lpp-2587.lzx test/bugs/lpp-2587.lzx >> svn add test/bugs/lpp-2587.lzx >> svn merge -r 1837:1838 http://svn.openlaszlo.org/openlaszlo/trunk/ >> test/bugs/lpp-2587.lzx test/bugs/lpp-2587.lzx >> >> And I need to resubmit my changeset. >> >> The reason was that for some reason test/bugs/lpp-2587.lzx hadn't >> been merged over from legals. It was created in r1719 and modified >> twice, in r1765 and r1838. The previous trunk->legals merge should >> have brought over that file in r1719 and merged the change from >> r1765, but it didn't. >> >> So this was an expedient -- and actually kind of a weird mix of >> expedience and caution. Why didn't I do: >> >> svn export -r 1719 http://svn.openlaszlo.org/openlaszlo/trunk/test/ >> bugs/lpp-2587.lzx test/bugs/lpp-2587.lzx >> svn merge -r 1764:1765 http://svn.openlaszlo.org/openlaszlo/trunk/ >> test/bugs/lpp-2587.lzx test/bugs/lpp-2587.lzx >> svn merge -r 1837:1838 http://svn.openlaszlo.org/openlaszlo/trunk/ >> test/bugs/lpp-2587.lzx test/bugs/lpp-2587.lzx >> >> to preserve change history? >> >> Or, if history wasn't a concern, simply >> >> svn export -r 1837 http://svn.openlaszlo.org/openlaszlo/trunk/test/ >> bugs/lpp-2587.lzx test/bugs/lpp-2587.lzx >> >> ? >> >> Anyway, I'm not sure it really matters. >> >> But we should look into why lpp-2587.lzx didn't make it into the >> last merge (which did cover 1719 and 1765). Perhaps other files >> didn't make it? I suppose I'll spelunk a bit on that. >> >> jim >> >> On Sep 19, 2006, at 5:48 PM, Benjamin Shine wrote: >> >>> Why did you have to do special handling of lpp-2587.lzx? >>> >>> On Sep 19, 2006, at 4:07 PM, Jim Grandy wrote: >>>> svn export -r 1765 http://svn.openlaszlo.org/openlaszlo/trunk/ >>>> test/bugs/lpp-2587.lzx test/bugs/lpp-2587.lzx >>> >>> >>> >> > _______________________________________________ Laszlo-dev mailing list [email protected] http://www.openlaszlo.org/mailman/listinfo/laszlo-dev
