-## A few bits about the RCS backends
+A few bits about the RCS backends
-### Terminology
+[[toc ]]
+
+## Terminology
``web-edit'' means that a page is edited by using the web (CGI) interface
as opposed to using a editor and the RCS interface.
-### [[Subversion]]
+## [[Subversion]]
Subversion was the first RCS to be supported by ikiwiki.
-#### How does it work internally?
+### How does it work internally?
Master repository M.
CGI operates on W. rcs_commit() will commit from W to M.
+For all the gory details of how ikiwiki handles this behind the scenes,
+see [[commit-internals]].
+
You browse and web-edit the wiki on W.
+W "belongs" to ikiwiki and should not be edited directly.
-### [darcs](http://darcs.net/) (not yet included)
+
+## [darcs](http://darcs.net/) (not yet included)
Support for using darcs as a backend is being worked on by [Thomas
-Schwinge](mailto:tschwinge@gnu.org).
+Schwinge](mailto:tschwinge@gnu.org), although development is on hold curretly.
+There is a patch in [[todo/darcs]].
-#### How will it work internally?
+### How will it work internally?
``Master'' repository R1.
CGI operates on R2. rcs_commit() will push from R2 to R1.
You browse the wiki on R1 and web-edit it on R2. This means for example
-that R2 needs to be updated from R1 if you are going the web-edit a page,
+that R2 needs to be updated from R1 if you are going to web-edit a page,
as the user otherwise might be irritated otherwise...
How do changes get from R1 to R2? Currently only internally in
-rcs_commit(). Is rcs_prepedit() suitable?
+rcs\_commit(). Is rcs\_prepedit() suitable?
It follows that the HTML rendering and the CGI handling can be completely
separated parts in ikiwiki.
What repository should [[RecentChanges]] and [[History]] work on? R1?
-##### Rationale for doing it differently than in the Subversion case
+#### Rationale for doing it differently than in the Subversion case
darcs is a distributed RCS, which means that every checkout of a
repository is equal to the repository it was checked-out from. There is
no forced hierarchy.
-R1 is the nevertheless called the master repository. It's used for
+R1 is nevertheless called the master repository. It's used for
collecting all the changes and publishing them: on the one hand via the
rendered HTML and on the other via the standard darcs RCS interface.
-R2, the repository where CGI operates on, is just a checkout of R1 and
+R2, the repository the CGI operates on, is just a checkout of R1 and
doesn't really differ from the other checkouts that people will branch
off from R1.
(To be continued.)
+
+#### Another possible approach
+
+Here's what I (tuomov) think, would be a “cleaner” approach:
+
+ 1. Upon starting to edit, Ikiwiki gets a copy of the page, and `darcs changes --context`.
+ This context _and_ the present version of the page are stored in as the “version” of the
+ page in a hidden control of the HTML.
+ Thus the HTML includes all that is needed to generate a patch wrt. to the state of the
+ repository at the time the edit was started. This is of course all that darcs needs.
+ 2. Once the user is done with editing, _Ikiwiki generates a patch bundle_ for darcs.
+ This should be easy with existing `Text::Diff` or somesuch modules, as the Web edits
+ only concern single files. The reason why the old version of the page is stored in
+ the HTML (possibly compressed) is that the diff can be generated.
+ 3. Now this patch bundle is applied with `darcs apply`, or sent by email for moderation…
+ there are many possibilities.
+
+This approach avoids some of the problems of concurrent edits that the previous one may have,
+although there may be conflicts, which may or may not propagate to the displayed web page.
+(Unfortunately there is not an option to `darcs apply` to generate some sort of ‘confliction resolution
+bundle’.) Also, only one repository is needed, as it is never directly modified
+by Ikiwiki.
+
+This approach might be applicable to other distributed VCSs as well, although they're not as oriented
+towards transmitting changes with standalone patch bundles (often by email) as darcs is.
+
+> The mercurial plugin seems to just use one repo and edit it directly - is
+> there some reason that's okay there but not for darcs? I agree with tuomov
+> that having just the one repo would be preferable; the point of a dvcs is
+> that there's no difference between one repo and another. I've got a
+> darcs.pm based on mercurial.pm, that's almost usable... --bma
+
+>> IMHO it comes down to whatever works well for a given RCS. Seems like
+>> the darcs approach _could_ be done with most any distributed system, but
+>> it might be overkill for some (or all?) While there is the incomplete darcs
+>> plugin in [[todo/darcs]], if you submit one that's complete, I will
+>> probably accept it into ikiwiki.. --[[Joey]]
+
+## [[Git]]
+
+Regarding the Git support, Recai says:
+
+I have been testing it for the past few days and it seems satisfactory. I
+haven't observed any race condition regarding the concurrent blog commits
+and it handles merge conflicts gracefully as far as I can see.
+
+As you may notice from the patch size, GIT support is not so trivial to
+implement (for me, at least). Being a fairly fresh code base it has some
+bugs. It also has some drawbacks (especially wrt merge which was the hard
+part). GIT doesn't have a similar functionality like 'svn merge -rOLD:NEW
+FILE' (please see the relevant comment in mergepast for more details), so I
+had to invent an ugly hack just for the purpose.
+
+By design, Git backend uses a "master-clone" repository pair approach in contrast
+to the single repository approach (here, _clone_ may be considered as the working
+copy of a fictious web user). Even though a single repository implementation is
+possible, it somewhat increases the code complexity of backend (I couldn't figure
+out a uniform method which doesn't depend on the prefered repository model, yet).
+By exploiting the fact that the master repo and _web user_'s repo (`srcdir`) are all
+on the same local machine, I suggest to create the latter with the "`git clone -l -s`"
+command to save disk space.
+
+Note that, as a rule of thumb, you should always put the rcs wrapper (`post-update`)
+into the master repository (`.git/hooks/`) as can be noticed in the Git wrappers of
+the sample [[ikiwiki.setup]].
+
+## [[Mercurial]]
+
+The Mercurial backend is still in a early phase, so it may not be mature
+enough, but it should be simple to understand and use.
+
+As Mercurial is a distributed RCS, it lacks the distinction between
+repository and working copy (every wc is a repo).
+
+This means that the Mercurial backend uses directly the repository as
+working copy (the master M and the working copy W described in the svn
+example are the same thing).
+
+You only need to specify 'srcdir' (the repository M) and 'destdir' (where
+the HTML will be generated).
+
+Master repository M.
+
+RCS commit from the outside are installed into M.
+
+M is directly used as working copy (M is also W).
+
+HTML is generated from the working copy in M. rcs_update() will update
+to the last committed revision in M (the same as 'hg update').
+If you use an 'update' hook you can generate automatically the HTML
+in the destination directory each time 'hg update' is called.
+
+CGI operates on M. rcs_commit() will commit directly in M.
+
+If you have any question or suggestion about the Mercurial backend
+please refer to [Emanuele](http://nerd.ocracy.org/em/)
+
+## [[tla]]
+
+## [Monotone](http://monotone.ca/)
+
+There is an unfinished patch in [[bugs/Monotone_rcs_support]].
+
+In normal use, monotone has a local database as well as a workspace/working copy.
+In ikiwiki terms, the local database takes the role of the master repository, and
+the srcdir is the workspace. As all monotone workspaces point to a default
+database, there is no need to tell ikiwiki explicitly about the "master" database. It
+will know. (BTW - this is also true of subversion. It might be possible to simplify the svn config?)
+
+The patch currently supports normal committing and getting the history of the page.
+To understand the parallel commit approach, you need to understand monotone's
+approach to conflicts:
+
+Monotone allows multiple micro-branches in the database. There is a command,
+`mtn merge`, that takes the heads of all these branches and merges them back together
+(turning the tree of branches into a dag). Conflicts in monotone (at time of writing)
+need to be resolved interactively during this merge process.
+It is important to note that having multiple heads is not an error condition in a
+monotone database. This condition will occur in normal use. In this case
+'update' will choose a head if it can, or complain and tell the user to merge.
+
+For the ikiwiki plugin, the monotone ikiwiki plugin borrows some ideas from the svn ikiwiki plugin.
+On prepedit() we record the revision that this change is based on (I'll refer to this as the prepedit revision). When the web user
+saves the page, we check if that is still the current revision. If it is, then we commit.
+If it isn't then we check to see if there were any changes by anyone else to the file
+we're editing while we've been editing (a diff bewteen the prepedit revision and the current rev).
+If there were no changes to the file we're editing then we commit as normal.
+All of this should work with the current patch.
+
+It is only if there have been parallel changes to the file we're trying to commit that
+things get hairy. In this case the current (implemented but untested) approach is to
+commit the web changes as a branch from the prepedit revision. This
+will leave the repository with multiple heads. At this stage, all data is saved, but there
+is no way to resolve the potential conflict using the web interface.
+
+In the specific case of a branch caused by a web edit, it may be possible to
+make monotone use the current web interface. This may be possible because we
+know that merging between the two revisions we have (the new branch
+and the prepedit revision) involves at most one conflicted file.
+We could use `mtn explicit_merge` to merge the revisions. If that
+succeeds without conflicts then good. If that fails, then we could
+use a special lua merge hook to spit out the conflict marked file
+and hand it back to the web interface and then abort the merge. At the same time, we'd have
+to modify the 'prepedit' data to include both parents so that when
+the user saves again we know we're in this case.
+
+If you get a commit and your prepedit data includes two revids then
+we form a commit manually using the automate interface - same way
+we currently build the micro-branch. However, while conflicts were being resolved,
+someone could have come
+along and introduced *another* one. So after forming this merge revision,
+you need to go back and check to see if the workspace revision has changed
+and possibly go through the whole process again. The repeats until you're
+merged.
+
+The end result of all of this is a system that can resolve all web conflicts without race
+conditions. (And because of the way monotone works it saves all data, including
+both sides of the merge, before the merge. You can go back later and check that
+the merge was reasonable.) It still doesn't provide a web-based way of merging multiple
+heads that come in through non-web interaction with monotone.