On Sat, Sep 03, 2005 at 05:04:23PM +0200, Daniel Lutz wrote:
> An other idea (for improving efficiency in a second step if necessary):
> Supposed some locking mechanism is implemented. Then caching could be
> introduced (user = e.g. script):
> 
> 1)  user requests write lock on the repository
> 2a) user requests file content of file A after patch X to stdout
>      -> generate a cache file (e.g. containing file A after patch X)
> 2b) user requests file content of file A after patch Y (after X) to 
> stdout
>      -> use cache file
> ...
> 3) user releases write lock
>      -> cache files are deleted

Yeah, that would be an neat trick, especially since the lock command would
be useful even in the absence of a caching mechanism.  Note that I'm still
leaning towards a lock command that releases the lock on exit.  If you want
to familiarize yourself with darcs and haskell, this would be a pretty easy
command to implement.

The procedure for creating a new command is pretty simple... I was about to
outline them here, but instead realized it would be smarter to add this to
the wiki, so you can pick up some advice at:

http://wiki.darcs.net/index.html/AddACommand

You could take any of the commands that create a lock as a starting point,
and pretty much just follow the instructions and gut the command (maybe
make it just putStr an instruction to hit return to release the lock and
then getChar and exit.
-- 
David Roundy
http://www.darcs.net

_______________________________________________
darcs-devel mailing list
[email protected]
http://www.abridgegame.org/cgi-bin/mailman/listinfo/darcs-devel

Reply via email to