[darcs-users] reviewing and testing pulls

[David Roundy]

On Mon, Apr 26, 2004 at 11:17:31PM -0700, Kevin Smith wrote:
> Andrew Pimlott wrote:
> >An important aspect for me of a revision control system is insight into,
> >and control over, changes.  It's not enough that I can reliably
> >incorporate changes from other trees, I want to be able to understand,
> >review, test, and possibly reject those changes.  
> 
> > 1a. How do see what a pull would give me, without doing the pull?
> 
> I share this view, and also find darcs weak in this area right now. This 
> is very similar to my posting a few days ago.
> 
> Most important to me is having confidence that it is merely a missing 
> feature, and not a structural limitation. If we can never have this 
> ability, then darcs may not be the tool for me.

It's certainly not a structural limitation.

> It seems like it would be quite possible to write a tool that would 
> compare your recorded repo against (the working directory + a patch 
> set). But what I really want is the ability to use any existing tool to 
> do that comparison.
> 
> Also, I want to build and test a "candidate" that includes an external 
> patch, prior to committing to including that patch in my repo. It seems 
> like a broken process if I have to put it in my repo first, and then 
> unpull or unrecord or rollback if I don't like it.

I'd say that the way to do this is to use a temporary repo, as Tommy and
Zooko suggest.  Using darcs get on a filesystem supporting hard links gives
you a temporary repo that costs in disk space only twice the contents of
the repository, and has the advantage of being a "candidate" in which you
can use the full power of darcs, recording new patches, unpulling some of
them, or pulling patches from multiple sources.
-- 
David Roundy
http://www.abridgegame.org