summaryrefslogtreecommitdiff
path: root/doc
diff options
context:
space:
mode:
authorjoey <joey@0fa5a96a-9a0e-0410-b3b2-a0fd24251071>2007-08-18 03:16:28 +0000
committerjoey <joey@0fa5a96a-9a0e-0410-b3b2-a0fd24251071>2007-08-18 03:16:28 +0000
commite30ba9ce7f9ead64901fb6f0a5764c6aff00f5f9 (patch)
treed6f03e9f4d80e4fd1ddf5638dd64bf1903a3d820 /doc
parent5f687c72b61550ef1fc20f438e5d66ce56e4f152 (diff)
web commit by http://willu.myopenid.com/: Update for new version of patch
Diffstat (limited to 'doc')
-rw-r--r--doc/bugs/Monotone_rcs_support.mdwn31
1 files changed, 7 insertions, 24 deletions
diff --git a/doc/bugs/Monotone_rcs_support.mdwn b/doc/bugs/Monotone_rcs_support.mdwn
index 95f759753..c976268f2 100644
--- a/doc/bugs/Monotone_rcs_support.mdwn
+++ b/doc/bugs/Monotone_rcs_support.mdwn
@@ -6,45 +6,28 @@ I've just made a patch to the ikiwiki code that allows it to use the [Monotone](
At the moment it is basically complete. At present rcs_notify() is implemeted but untested, the rest is implemented and tested.
-In an older version of the patch
-there was support for simultaneous commits, but any conflicts left the repository in an unmerged state which had to be cleaned up using the non-web
-monotone interface. The user's data was still stored, and the user was informed if that happened.
-
The current version of the patch handles conflicts through the web interface. It is still not perfect as it will break if there is a rename that conflicts with a web change (but so will the other Rcs plugins I think). It also commits a revision with conflict markers if there is a conflict requiring such markers... ick.
-Note: This patch requires the Monotone perl module, which is only available in Monotone 0.36 or later. The module is in the Monotone contrib/ directory. While 0.36 is relatively new, that Monotone module itself will work with a monotone binary back quite a few releases.
-Apparently this module did't make some versions of the monotone 0.36 release tarball. It is available from the monotone repository here: <http://viewmtn.angrygoats.net/branch/changes/net.venge.monotone>.
+Note: This patch requires a rather recent Monotone perl module (18 August 2007 or later). It is available from the monotone repository here: <http://viewmtn.angrygoats.net/branch/changes/net.venge.monotone>.
> The setup instructions to add 40 lines of code to monotonerc is pretty frightning stuff.
> Is there some way this can be automated? --[[Joey]]
->> I've committed a bunch of this to monotone so that in future it could be removed. I didn't
->> want to remove it from the instructions until it was in a monotone release though.
-
->>> Fair enough. Didn't realize you were a monotone committer. :-) As long as it goes away later I don't mind having it in setup initially.
+>> I've committed a bunch of this to monotone so that in future it could be removed.
+>> I've also just fixed this so it is in a separate, automagically generated, rc file.
->> The real issue here is that there is a mismatch between ikiwiki's conflict model and
->> monotone's. I need to get code into monotone to add conflict markers - that is
->> the code that goes in the monotonerc. In practice I could add this to any file I know
->> how to reference and I can then tell monotone about it with a command line arg.
->> Is there a good place for such a file?
+>>> Fair enough. Didn't realize you were a monotone committer. :-)
->>> There's not a really good place. /usr/share/ikiwiki/foo would do,
->>> except for the systems where /usr/share/ is elsewhere. It would need to be
->>> runtime configurable just like the basewiki and template locations are.
->>>
->>> Hmm, another option would be to put it in the wiki's .ikiwiki directory when
->>> ikiwiki starts up. This is the technique used by the search plugin to create
->>> various data and config files that hyperestradier needs.
+>>>> I am, but still a little newish. Feedback is good.
> Having rcs_commit return a warning message when there's an unresolved conflict
> isn't right; that message will populate the page edit box. You might want
> to use the error() function here?
->> It should never reach that case, so yes, I can change that to error.
+>> It should never reach that case, so I have changed that to error.
> There'an incomplete comment ending with "note, this relies on the fact that"
->> erg... sorry, will fix.
+>> erg... sorry, fixed.
[[tag patch]]