[darcs-users] Patches dependency discovery

Radosław Grzanka radoslawg at gmail.com
Mon Jul 11 07:06:18 UTC 2005


On 7/9/05, David Roundy <droundy at darcs.net> wrote:
> On Fri, Jul 08, 2005 at 06:57:46PM +0200, Tristan Seligmann wrote:

> Another way to look at this (which I think is equivalent) is that there are
> two states for each patch.  One is the tristate, reflecting whether the
> user has explicitely said yes or no to that patch.  The other is the
> "current" state of the patch--whether it's on the to-be-applied (or
> to-be-unapplied, etc) list.  This second state is a bi-state, and somehow
> needs to be displayed along with the did-the-user-choose-this-state state.
> I.e. there will be four possible states.  Either the user wants the patch,
> she doesn't want the patch, or she hasn't said, but it's required by one
> that she wants, or she hasn't said and it's not required by one that she
> wants.  I suppose there's one more state, which is that she hasn't said,
> but it's forbidden by a patch that she's said she doesn't want.

I like the idea of forbidding some patches but I am not sure whether
this is not an overkill.. I have mixed feeling whether it is worth the
effort.
The problem I see is mainly in UI. How to support this? Another
checkbox which says: Forbidden? Or right click and 'Forbid' Action?
The first makes IMO UI a bit unreadable (and we don't have four state
checkboxes.. don't we?). The latter may be non intuitive.

> 
> And just to reiterate what I've suggested in a previous thread, I think if
> you want a "flat list with interesting states" sort of interface, this can
> be relatively easily acheived with a pipe-oriented extension to the current
> interactive interface, which would allow eclipse to open up a bidirectional
> pipe to darcs and inform darcs which patches have been selected and query
> the states of the other patches.  And in the process we'd also improve the
> interactive interface, which would be nice.

This may be done but It requires some having thought and testing.

BTW. Does darcs 'lock' the repository when user runs pull action
(during interactive period)? Just the random thought.

> By the way, the semi-cruddy gui interface uses simple two-state buttons,
> and even with that it is a huge improvement as a user interface, being able
> to select a patch and watch the required patches get selected... I mean to
> say that I was surprised at how much nicer an experience it was to use.
> Which is why I haven't been convinced to remove that code, even though it's
> pretty darn crufty at this stage.

Agree!  :D

Cheers,
  Radek.

-- 
Galiera: http://zdjecia.zuzia.homelinux.net/
Gallery: http://zdjecia.zuzia.homelinux.net/




More information about the darcs-users mailing list