Hello,
A Haskeller by the name of John Macfarlane has created a nice wik engine in
Haskell on top of Happs and a distributed version control system.  You can
see it here or read the original release announcement:
http://johnmacfarlane.net:5001/
http://www.haskell.org/pipermail/haskell-cafe/2008-November/050336.html

Now, it just so happens that the backend isn't Darcs, but I thought I read
somewhere that the author is open to having a darcs backend if we provide a
programmatic way to invoke darcs (aka libdarcs).

Given that the wiki is using a vcs in the background it is possible to
author pages using just the version control backend.

Also it supports code snippets with syntax highlighting and LaTeX math stuff
and exports to tons of formats (including latex, meaning PDF output probably
isn't a hard feature to add if it could be done server side from the latex
output).  Internally it uses Pandoc making the input and output
possibilities quite numerable.  reST is even an option; but it seems
markdown is the default input type.

So, this is me publicly suggesting that we consider, or evaluate, the
potential use of gitit for the darcs project.  Here are some of the reasons
why I think this is a good idea:
1) Assuming that the darcs project is user #0 of libdarcs this could be used
to give us user #1.
2) Trent has mentioned he would like a wiki engine that can be used without
a web browser on multiple occasions.
3) With support for both Haskell code and math we could potentially move the
entire manual, including patch theory sections, to the wiki reducing the
barrier to documentation authoring beyond the current threshold.
4) We could script the creation of the pdf manual from the wiki content,
same with the manpages.
5) In my opinion the existing feature set of gitit makes it superior already
to what we are currently using.

Potential problems:
a) gitit is still young.
b) libdarcs isn't ready *today* to be the backend.
c) Porting the existing wiki content.

For (a) maybe we should play it safe and wait X months?  How about X = 3-6
to see if gitit needs to stabilize?  For (b) I'd say git is popular and
respectable, but I think we should dogfood our own vcs whenever we can
because of the message that sends.  This means we need to figure out how to
get libdarcs to work for gitit and find out if John would be okay
maintaining a backend layer for libdarcs.  I bet we could find a way to
automate, or mostly, automate (c).

What do others think?

Thanks!
Jason
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to