[darcs-users] darcs conflicts/dependencies -- is patch theory the place to start?

[AntC]

Gracjan Polak <gracjanpolak <at> gmail.com> writes:

> 
> 2012/9/13 Stephen J. Turnbull <stephen <at> xemacs.org>
> 
> The problem is identifying weakest preconditions.
> 
> 
> This got me interested. Is there a paper/odc/wiki where it is described how 
hard does darcs right now try to identifying weakest preconditions?
>

Hi Gracjan, I think we need to start earlier than that. I've not seen (perhaps 
I didn't look in the right places) any detailed explanation of the conditions 
that cause a conflict.

It's described informally as where you can't commute two patches, and we can 
kinda guess what that means.

What's nice about the L/S/L model is they have a very precise definition of 
pre-conditions (and post-conditions -- since these are just the pre-conds when 
you invert; and invariants -- these are both pre- and post-). And they can 
describe it in terms of set membership (presence/absence in the repo).

My post about the components of a primitive patch (action, location, content) 
was working towards it. (A pre-cond on the location is that it must exist; for 
a create/add action the pre-cond is that the 'target' doesn't exist; etc)

AntC