On 7/17/2010 14:22, Eric Kow wrote:
On Sat, Jul 17, 2010 at 19:01:25 +0100, Ganesh Sittampalam wrote:
In March, there were three or four people using the darcs library, a
veritable explosion :-). With at least two of them, questions about
things like filtering lists arose. Things have gone a bit more quiet
again since then, but I do feel that nipping any dangerous usage in
the bud is a good thing.
Just seeing if this code making it to stable is itself urgent
or if by urgency we're mostly just talking about Darcs hackers.
In my view, there is some urgency in it getting into stable for the
library. For the executable, changing HEAD is fine.
In that case (and despite it being a way off before we'll get any
maintenance benefits), I guess I would somewhat uncharacteristically say
"punch it" (let's flip that switch as soon as you feel we can reasonably
do it)
(I might even go so far as to say that we should go ahead and flip
both switches at the same time if that makes life any simpler)
Considering that the library users have already been warned that the
library is somewhat unstable, I think it is a smart move to go ahead and
switch the library to using witnesses.
On the other hand, I don't think that 2.5 is the time to switch
executables over to witness builds just yet. I think most of us would
hate for some unexpected side-effect of witnesses builds affect
day-to-day usage of a 2.5 executable. Not that I think there would be
any, but letting library users and HEAD users test such a big change
first sounds to me preferable to a "surprise witnesses attack". At the
very least a comparable witness build and non-witness build should be
benchmarked.
--
--Max Battcher--
http://worldmaker.net
_______________________________________________
darcs-users mailing list
darcs-users@darcs.net
http://lists.osuosl.org/mailman/listinfo/darcs-users
