[darcs-users] darcs convert on non-independent repos

[Ganesh Sittampalam]

Hi,

'darcs convert' claims the following restriction:

> Furthermore, darcs 2 repositories created by different invocations of 
> this command SHOULD NOT exchange patches, unless those repositories had 
> no patches in common when they were converted.  (That is, within a set 
> of repos that exchange patches, no patch should be converted more than 
> once.)

I think this condition is actually a little too restrictive, which makes 
sense in terms of providing simple and comprehensible documentation.

In particular, I believe the following scenario is also OK:

repo X = A;B
repo Y = A;C

where A, B and C are all sequences of patches, A is in the same order in 
both X and Y, and B and C have no patches at all in common.

In other words you can convert both X and Y, and then exchange patches 
between the converted repos without any problems.

The reason I think this is ok is that darcs convert acts on the current 
representation of any given patch, which makes its behaviour dependent on 
the current order of patches in a repo, but other than that its behaviour 
on a particular patch is independent of the other patches in the repo - 
it's just a map operation over the current patch sequence.

So in the above scenario, we should get identical results for A, and 
independent results for B and C.

Can anyone pick any holes in this reasoning?

(I'm hoping that it'll offer a nice way out of Axel Simon's current 
problem with gtk2hs)

Cheers,

Ganesh