[darcs-users] About darcs darcs repo as an darcs usage example...

Nicolas Pouillard nicolas.pouillard at gmail.com
Tue Mar 10 10:30:00 UTC 2009


Excerpts from Petr Rockai's message of Tue Mar 10 10:13:37 +0100 2009:
> Hi,
> 
> every now and then, someone mentions the size of the darcs darcs repo as the
> measure of where darcs scales. I'd like to warn people that before they do
> that, they should mention, that darcs darcs repo is anxiously avoiding any of
> the more troublesome aspects of darcs. Eg. there are very few conflict
> resolutions in darcs darcs repo. We impose a very strict workflow on our
> contributors, which very few projects will be willing to adopt (even in darcs
> itself, there's a fair amount of dissent). Other workflows will result in many
> more conflict resolutions, sometimes in conflict fights and repos of size of
> the darcs repo could wind up unusable due to merging slowness and bugs (we
> tried to branch darcs repo with anything other than minor changes twice, both
> attempts resulting in massive problems with merging and ultimately loss of the
> branch due to its un-usability).
> 
> We should keep our marketing honest. I wouldn't recommend using darcs to any
> medium project (ie. more than a couple hundred patches and a single, at most
> two contributors) without a darcs expert on board. At least not until the core
> is either rewritten or debugged. To put it simply, while darcs is smart, it's
> not reliable, and in an RCS, that's a catastrophe waiting to happen.

On the other end, people that follow the strict workflow of avoiding conflicts
as much as possible, can reasonably have a repo of the size of darcs itself.
The only long-term restriction is to have only short-lived branches of the
synchronized trunk. So I would say that depending on how you treat conflicts,
darcs may scale from small projects (as you've described) to medium projects
(as darcs itself).

All the best,

-- 
Nicolas Pouillard


More information about the darcs-users mailing list