mercurial as storage for (big) wiki

[Tim Post]

On Thu, 2007-05-03 at 09:25 +0200, solo turn wrote:
> hi,
> 
> would it be possible to use mercurial as storage engine of a wiki,
> like mediawiki?
> 
> i.e. i guess the following should be matched:
> * store into database
> * full text search possible
> 
> this would allow to "transplant" changes into clean versions of wiki pages.
> 
> if mercurial is not suitable, is there any version control which is
> suitable for this - or is it better to add "branching" and "cherry
> picking" to standard wiki stores?
> 
> -solo

Your not the only one who sees interesting uses that are more 'side
benefits' of mercurial. I wish more PHP developers, theme authors and
hosting control panels took advantage of it as an installer. 

I'm afraid that binary 'diffing' would be better accomplished using some
sort of database. SQLite can and should be used much more than it is. 

For what you want, I suggest you adopt the 2.6.20.3 Linux kernel and
install the ext3cow file system. Store your databases on an ext3cow file
system. If you use many databases, that is many points in time you can
roll back independently in many places.

ext3cow does not depend on snapshots. You just roll back the clock to
any point in time, and that's the file system you see.

So every second that passes is another 'immutable' state of your file
system. Just mark points in time like you would hg commit, and roll
back / forward to make them what you want. I'm sure you could think of
ways to add other functionality too.

You can find out more about ext3cow at http://www.ext3cow.com

It would be a bit of work, would lend better to a clustered sql scenario
actually. I think it would do what you need though.

Man, the stuff I think up just to avoid really working on my python. I
tell ya.

Best,
--Tim

> _______________________________________________
> Mercurial mailing list
> Mercurial at selenic.com
> http://selenic.com/mailman/listinfo/mercurial