[darcs-users] proposal for 2.8 : fastconvert (and remark on the process)

Stephen J. Turnbull turnbull at sk.tsukuba.ac.jp
Wed Dec 1 02:37:37 UTC 2010


Petr Rockai writes:

 > It's not really a plugin, although if we model the plugin system after
 > git, it could be one, by the virtue of its name. I believe git foo
 > ... just calls git-foo ... without any special considerations.

Yes, it's basically just an exec() (maybe even a system()) (up to the
precedence of built-ins vs. callable commands, which I forget).

 > (Things work differently in hg, as far as I know.)

Yes, because the extension language is Python, not POSIX shell.

 > The process does sound a bit over-engineered to me,

Possibly.  But remember, all of the other modern VCSes disagree with
you.  They all provide a plug-in architecture, and they all use it to
great advantage in establishing best practice before adding new
features to the core.

 > and also like a lot of (extra) work

Perhaps.  Certainly the bzr people frequently tie themselves up in
knots over their plugins (but there the problem is that the knotty
plug-ins are really part of the core; you can hardly get anything
done, including unit testing, without them, but because of automatic
plug-in discovery, doing unit testing with plug-ins enabled is kinda
fraught with potential for confusion).


More information about the darcs-users mailing list