Moving to D

[Ulrik Mikaelsson]

2011/1/6 Nick Sabalausky <a at a.a>:
> "Caligo" <iteronvexor at gmail.com> wrote in message
> news:mailman.451.1294306555.4748.digitalmars-d at puremagic.com...
>> Perhaps using a modern SCM like Git might help?  Everyone could have (and
>> should have) commit rights, and they would send pull requests.  You or one
>> of the managers would then review the changes and pull and merge with the
>> main branch.  It works great; just checkout out Rubinius on Github to see
>> what I mean: https://github.com/evanphx/rubinius
>
> I'm not sure I see how that's any different from everyone having "create and
> submit a patch" rights, and then having Walter or one of the managers review
> the changes and merge/patch with the main branch.

With the risk of starting yet another VCS-flamewar: It gives the
downstream developers an easier option to work on multiple patches in
patch-sets. Many non-trivial changes are too big to do in a single
step, but requires series of changes. Sure, the downstream hacker
could maintain import/conversion to VCS, but with added job, and when
Walter or someone else gets to review they are no longer
well-annotated patches.

It also facilitates a setup where Walter (BDFL? ;) starts to trust
some contributors (if he wants to) more than others, for them to work
on private branches and submit larger series of patches for each
release.

Especially, when you detect a showstopper bug that blocks your
progress, IMHO, it's easier using a DVCS to maintain a local patch for
the needed fix, until upstream includes it. I've often used that
strategy both in D-related and other projects just to remain sane and
work-around upstream bugs, I just usually have to jump through a some
hoops getting the source into DVCS in the first place.

I think it was on this list I saw the comparison of VCS:es to the
Blub-problem? http://en.wikipedia.org/wiki/Blub#Blub

Although, I don't think the current setup have any _serious_ problems,
I think there might be slight advantages to gain. OTOH, unless other
current key contributors wants to push it, it's probably not worth the
cost of change.