Created on 2009-10-16.13:14:18 by mornfall, last changed 2009-10-21.11:13:43 by kowey.
| msg8961 (view) |
Author: mornfall |
Date: 2009-10-16.13:14:15 |
|
It turns out it's not quite possible to resolve a conflict by going to the
version before the conflict using standard constructs. First try: darcs rec: No
changes!. Another: darcs rec --ask: can't select both conflicting sides. The
only way out is darcs tag (that actually works, somewhat contrary to
intuition), but is still something I'd rather not do.
|
| msg8964 (view) |
Author: kowey |
Date: 2009-10-19.11:41:25 |
|
Details, please? :-)
|
| msg8969 (view) |
Author: mornfall |
Date: 2009-10-19.22:44:03 |
|
Get a conflict, resolve it by removing everything between vvv and ^^^ markers
(i.e. *both* sides of the conflict). But that is the pristine state, so no way
to record that as a resolution. You get the same effect from reverting. You
cannot record an empty patch depending on the two conflicting patches either,
since --ask-deps does not let you do that. So you end up recording a pair of
patches, one to make an arbitrary non-empty resolution and then roll it back.
Or so. Not-so-nice.
|
| msg8975 (view) |
Author: kowey |
Date: 2009-10-21.10:13:01 |
|
OK, I think I'm going to downgrade this to a feature (I'm flip-floppy on this,
but I doubt it matters very much).
It sounds like we'd need somebody to think of a way how this should work on the
UI level (hopefully without having to add any new constructs).
I *guess* one workaround might be to record the arbitrary non-empty resolution
(eg. the conflict markers), rollback, unrecord and amend the first patch. (I'm
sure there's a less convoluted way around it... but I think you get the idea).
Also: is it obvious why --ask-deps doesn't allow this?
|
| msg8976 (view) |
Author: mornfall |
Date: 2009-10-21.11:09:09 |
|
Eric Kow <bugs@darcs.net> writes:
This probably won't work, since the amended patch should become empty if
coalescing works, I guess?
|
| msg8977 (view) |
Author: kowey |
Date: 2009-10-21.11:13:22 |
|
On Wed, Oct 21, 2009 at 11:09:12 +0000, Petr Ročkai wrote:
> This probably won't work, since the amended patch should become empty if
> coalescing works, I guess?
Isn't making the amended patch empty the point?
|
| msg8978 (view) |
Author: kowey |
Date: 2009-10-21.11:13:40 |
|
On Wed, Oct 21, 2009 at 12:12:39 +0100, Eric Kow wrote:
> On Wed, Oct 21, 2009 at 11:09:12 +0000, Petr Ročkai wrote:
> > This probably won't work, since the amended patch should become empty if
> > coalescing works, I guess?
>
> Isn't making the amended patch empty the point?
Oh! nevermind :-)
|
|
| Date |
User |
Action |
Args |
| 2009-10-16 13:14:18 | mornfall | create | |
| 2009-10-19 11:41:29 | kowey | set | status: unknown -> waiting-for
nosy:
+ kowey
topic:
+ Conflicts
messages:
+ msg8964
assignedto: mornfall |
| 2009-10-19 22:44:07 | mornfall | set | nosy:
kowey, darcs-devel, dmitry.kurochkin, mornfall
messages:
+ msg8969 |
| 2009-10-21 10:13:05 | kowey | set | status: waiting-for -> needs-reproduction
title: impossible to resolve conflict by reversion -> null conflict resolutions
nosy:
kowey, darcs-devel, dmitry.kurochkin, mornfall
messages:
+ msg8975
priority: bug -> feature
assignedto: mornfall -> |
| 2009-10-21 11:09:12 | mornfall | set | nosy:
kowey, darcs-devel, dmitry.kurochkin, mornfall
messages:
+ msg8976 |
| 2009-10-21 11:13:24 | kowey | set | nosy:
kowey, darcs-devel, dmitry.kurochkin, mornfall
messages:
+ msg8977 |
| 2009-10-21 11:13:43 | kowey | set | nosy:
kowey, darcs-devel, dmitry.kurochkin, mornfall
messages:
+ msg8978 |
|
