[darcs-users] Darcs Wiki Engine

[Max Battcher]

Gwern Branwen wrote:
> I too have cast an avaricious eye over gitit. I am unsure that we need
> libdarcs, however. Take a look at the module which handles the actual
> Git integration for gitit:
> http://github.com/jgm/gitit/tree/master/Gitit%2FGit.hs
> 
> It is basically 10 functions which are not much more than
> shell-scripting. Most of them look like they could easily be replaced
> by Darcs equivalents. The only ones that look really different are the
> ones which use the SHA1 identifiers, and presumably for Darcs that
> would be replaced by simply the name of a patch.

No, you probably want to use the patch hashes (from darcs changes 
--xml-output; darcs x --match="hash ...").

I do think that all of these functions could be implemented with the 
existing command line interface, with perhaps a possible exception of 
gitGrep (use real grep instead? temporary repositories may be a concern?).

On the other hand, if people really are desperate for a libdarcs 
approach (git seems fine in this example without a libgit), I think this 
could serve as a good minimal case to begin with for a libdarcs0:

1) It just needs to export to Haskell code.
2) There is a very small subset of necessary API calls, most of which 
already exist at the CLI/command/subcommand level.
3) It will be immediately usable by an implementing third party 
application that can serve as the benchmark/proof-of-concept.

It seems better than "export everything and close/deprecate features as 
we see fit" to go "start with an existing app(s) and add features as 
feedback comes in".

Ultimately, however I think worrying about libdarcs in this case is a 
red herring: the first focus should be performance.  Given CLI commands 
that run nearly as fast as their git counterparts is probably more 
useful to everyone involved (even with shell-scripting 
interop/marshalling concerns).

--
--Max Battcher--
http://worldmaker.net