darcs

Issue 252 Check and repair should check the pristine tree: same time == same file

Title Check and repair should check the pristine tree: same time == same file
Priority feature Status resolved
Milestone Resolved in
Superseder Nosy List Aaron, darcs-devel, dmitry.kurochkin, jch, kowey, markstos, mornfall, thorkilnaur, tommy
Assigned To
Topics

Created on 2006-08-28.13:30:13 by lists2546, last changed 2010-01-19.01:22:44 by mornfall.

Messages
msg939 (view) Author: Aaron Date: 2006-08-28.13:30:05
I have a repository in which "darcs diff" correctly shows an unrecorded
modification, but "darcs whatsnew" says "No changes."  Darcs repair says
that the repo is consistent (nothing to repair).

When I copy the repo using cp -r, darcs whatsnew does find the change in
the new copy.  When I copy it using cp -rd, which preserves links, it
doesn't find it in the new repo.

I think I got the repo into this state by the following steps.  I
recorded patch 1, followed by patch 2, which both change the same file. 
Then I noticed a bug.  I made a modification in the file, tested it, and
decided to amend-record patch 1 with this modification.  I first had to
unpull patch 2, because 2 depended on 1 so darcs wouldn't let me amend 1
while 2 was in the repo.  Then I ran amend-record, and even though the
modification was still there in the file, darcs didn't find any changes
to amend with.

-Aaron
msg944 (view) Author: jch Date: 2006-08-31.21:56:37
What does ``darcs what --ignore-times'' do?
msg955 (view) Author: Aaron Date: 2006-09-04.07:22:55
OK, with --ignore-times it shows the change correctly.  I don't
understand how the series of actions I described got it into a state
with incorrect timestamps, but maybe it didn't happen exactly as I
described.
msg956 (view) Author: jch Date: 2006-09-04.07:41:11
I guess the actual issue is that darcs repair didn't notice there was a
problem.

Retitling bug, changing priority.  Thanks for your help, Aaron.

Juliusz
msg963 (view) Author: droundy Date: 2006-09-05.20:19:46
On Mon, Sep 04, 2006 at 07:41:25AM +0000, Juliusz Chroboczek wrote:
> I guess the actual issue is that darcs repair didn't notice there
> was a problem.

I see, that makes sense.  I suppose we'd also want darcs check to
check that files with same timestamp in pristine and the working
directories are actually identical.
-- 
David Roundy
msg3278 (view) Author: markstos Date: 2008-02-09.19:02:03
I believe this is considered "resolved-in-unstable" when using the --hashed and
--darcs-2 formats, which no longer store the pristine as plain files, but in a
special hashed format.
msg3446 (view) Author: droundy Date: 2008-02-15.15:19:07
Actually, this isn't resolved, since we still use timestamps to avoid reading
the contents of files in the working directory.  :(
msg7003 (view) Author: thorkilnaur Date: 2009-01-06.17:23:59
As this is considered a new feature, I set status deferred.

Best regards
Thorkil
msg7073 (view) Author: thorkilnaur Date: 2009-01-13.11:04:37
Sorry, deferring was a mistake. Instead, I'll put it into the need-info state, 
asking for a discussion of whether we should work towards supporting this 
feature.

Thanks and best regards
Thorkil
msg8721 (view) Author: kowey Date: 2009-09-06.20:48:43
Worth doing.

I'll just tack Petr on the nosy list for this because I think he may want this
want called to his attention.
msg8728 (view) Author: mornfall Date: 2009-09-07.08:15:24
Well, the hashed-storage index changes the situation here. Moreover, darcs-hs,
when doing check, makes sure that the index is OK (i.e. that the data in the
index actually matches the data in the working tree) in this regard, so when
the darcs-hs branch is merged, it will just be a matter of fixing up repair,
which is trivial.
msg9857 (view) Author: mornfall Date: 2010-01-19.01:22:42
This is fixed now (will be part of 2.4). That is, check will detect and repair 
will correct any discrepancies between what has really changed and what darcs 
thinks has changed. Moreover, it now should be harder to run into the original 
problem due to the way index caches the timestamps/hashes. At least in theory.
History
Date User Action Args
2006-08-28 13:30:13lists2546create
2006-08-31 21:56:41jchsetstatus: unread -> unknown
nosy: + jch
messages: + msg944
2006-08-31 23:00:06jchsetstatus: unknown -> waiting-for
2006-09-04 07:23:02lists2546setmessages: + msg955
2006-09-04 07:41:25jchsetpriority: bug -> feature
messages: + msg956
title: Difference between darcs whatsnew and darcs diff -> Check and repair should check the pristine tree for inconsistent times
2006-09-05 20:19:53droundysetmessages: + msg963
2007-03-09 10:27:35Aaronsetnosy: + Aaron, kowey, beschmi
2007-03-09 10:29:06Aaronsetnosy: - lists2546
2007-08-01 07:00:56koweysetstatus: waiting-for -> unknown
nosy: droundy, jch, tommy, beschmi, kowey, Aaron
2008-02-09 19:02:04markstossetstatus: unknown -> resolved-in-unstable
nosy: + markstos
messages: + msg3278
2008-02-15 15:19:08droundysetstatus: resolved-in-unstable -> deferred
nosy: droundy, jch, tommy, beschmi, kowey, markstos, Aaron
messages: + msg3446
2008-05-07 15:27:04droundysetstatus: deferred -> unknown
nosy: + dagit
2009-01-06 17:24:02thorkilnaursetstatus: unknown -> deferred
nosy: + dmitry.kurochkin, simon, thorkilnaur
messages: + msg7003
2009-01-13 11:04:40thorkilnaursetstatus: deferred -> waiting-for
nosy: droundy, jch, tommy, beschmi, kowey, markstos, dagit, simon, Aaron, thorkilnaur, dmitry.kurochkin
messages: + msg7073
2009-08-06 17:35:07adminsetnosy: + jast, Serware, darcs-devel, zooko, mornfall, - droundy, jch, Aaron
2009-08-06 20:32:17adminsetnosy: - beschmi
2009-08-10 21:54:58adminsetnosy: + Aaron, jch, - darcs-devel, zooko, jast, Serware, mornfall
2009-08-10 23:55:52adminsetnosy: - dagit
2009-08-25 17:38:18adminsetnosy: + darcs-devel, - simon
2009-08-27 14:14:49adminsetnosy: jch, tommy, kowey, markstos, darcs-devel, Aaron, thorkilnaur, dmitry.kurochkin
2009-09-06 20:48:45koweysetstatus: waiting-for -> needs-implementation
nosy: + mornfall
messages: + msg8721
title: Check and repair should check the pristine tree for inconsistent times -> Check and repair should check the pristine tree: same time == same file
2009-09-07 08:15:27mornfallsetstatus: needs-implementation -> needs-reproduction
nosy: jch, tommy, kowey, markstos, darcs-devel, Aaron, thorkilnaur, dmitry.kurochkin, mornfall
messages: + msg8728
2010-01-19 01:22:44mornfallsetstatus: needs-reproduction -> resolved
messages: + msg9857