David Roundy wrote:
[...]
This is a bit odd. Is this a repository that you can make available?
I've temporarily placed it at:
http://members.dodo.com.au/~netocrat/fish
And
do you still have the two repositories so this can be reproduced?
I can reproduce it using one repository:
darcs get http://members.dodo.com.au/~netocrat/fish
cd fish
darcs unpull
[select "y" to unpull "Resolution of conflicts with selective array
conversion patch"]
darcs unpull
[select "y" to unpull "Selective colon<->array conversions"]
Now re-pull the first patch only ("Selective colon<->array conversions")
to have the conflicts marked (I have no prefs/defaults settings file):
darcs pull
[select "y" then "n"]
less env.c
[shows the conflict including the do_arraysep_conv line]
To see that the external merge omits this line:
darcs revert
[select "y" continuously to revert all changes]
darcs unpull
[select "y" to unpull "Selective colon<->array conversions"]
darcs pull --external-merge 'kdiff3 --output %o %a %1 %2'
[select "y" then "n"]
[shows the three versions of less.c, none of which have do_arraysep_conv
listed in the same place as with the internal conflict resolution
approach above]
I'm a little hazy on how the repository got to the state it's in - there
was some degree of amended-record and out-of-order pulling/unpulling
going on, but if pressed I could probably figure it out.
My best guess is that you've got a three-way conflict here (if this isn't a
bug). External resolution doesn't work well for worse than two-way
conflicts.
But should at least present information consistent with the internal
resolution approach, right?
--
http://members.dodo.com.au/~netocrat
_______________________________________________
darcs-users mailing list
[email protected]
http://www.abridgegame.org/mailman/listinfo/darcs-users
