[darcs-users] GSoC wrap-up mode

Petr Rockai me at mornfall.net
Mon Aug 3 16:53:39 UTC 2009


Jason Dagit <dagit at codersbase.com> writes:
> -1 to a quick 2.4.  Lots has changed, right?  I say we give that a chance to be
> tested in the real-world a bit.  I think we should be prudent here.  Your work
> is more than simple refactors and bug fixes as I understand it.  You wrote a
> lot of new code/abstractions.  So I think the prudent thing is to dog food that
> for a while till we have internal confidence, then push it out to the world.

That's relative. Most of the code is currently employed by 2.3 in the whatsnew
path. There's very little other "new" stuff in darcs-hs -- bulk of this is just
one-liners flipping things from unsafeDiff to treeDiff.

As for internal confidence, where would that come from? I mean, I am using
darcs-hs as my main darcs for over a month now. Daily usage for two months or
five, by one or three people is not much of a difference. My argument here is,
that all that extra time will cause is sloppy review, since people will say
"oh, but we have 5 months to test this, so we don't need to pay so much
attention". I reckon that the net result with longer schedule will be worse
than with short schedule. Don't worry, I don't mean to rush things. But I would
like to work further, and I feel that lumping twice as many things in a single
release is not going to be any easier. However, waiting six months to get
things into mainline because "we already have lots of things for 2.4 that's due
early next year" would be probably quite demotivational.

>     There are relative merits to both approaches. What I don't like about
>     waiting six months for integrating darcs-hs into a stable release is that
>     it gives us too much slack time. We will put things off because we can
>     and then do a poor QA job, just because the thing will be old by then.
>
> Random thought:
> I say we should have an unstable branch and a stable branch :)  Leave darcs-hs
> out of stable for now, and keep working on and testing the unstable  branch.  I
> doubt this will be popular, but it seems reasonable to me considering we seem
> to have two competing goals.  It seems like there is 1) the goal of providing
> the same stable darcs that we have been providing; 2) the goal of stablizing
> darcs-hs. 

Likewise, I think that'll just provide us with more excuses. We either are
confident in the code, or we are not. I don't buy the argument that waiting
arbitrary number of months is going to make us any more confident in the
code. We have been collectively using darcs-2 core for what, a year now? Yet, I
am not any more confident about it than I was at the beginning. I would rather
see people actually taking action and poking the code actively wherever they
don't feel confident about it. But if they don't do that if given 2 months,
will they if given 5? Hard to tell, but my bet would be on "no".

(We know how to poke at darcs-2 core to make it fall apart. But unlike the
darcs-2 core, if you find ways to break hashed-storage, I will go and fix it so
that it won't, which will translate into a better release.)

Yours,
   Petr.


More information about the darcs-users mailing list