Version control systems

[Isaac Dupree]

Duncan Coutts wrote:
> On Sat, 2008-08-09 at 15:46 +1000, Manuel M T Chakravarty wrote:
> 
>> Raising the bar for developers to contribute to a project has been  
>> proven to be a very bad idea many times.  Let's not take GHC down that  
>> path.
> 
> I don't especially relish having to learn another vcs tool or raising
> the bar for contributions to Cabal either (we have lots of people who
> make small one-off contributions).

I wonder how many of the libraries are "core" in that they 
need to be changed a lot for GHC?
- all the ones that depend on GHC internals, such as base. 
(Except the current system has many of them use preprocessor 
conditionals so that can they depend on various compilers' 
internals, including nhc98 and hugs?  Because a lot of that 
code is actually shared between implementations)
- Cabal, since it's needing a lot of extension to make GHC 
work with it.

Do boot-libraries like unix typically need work by GHC devs?

On the other hand, it's looking like there's enough 
intersection between GHC and other-haskell that it's not 
such a helpful path to pursue.

not quite related: I wonder about various haskell libs 
switching to darcs2 format.  A few new programs use it 
already.  As distros include darcs2, it should become less 
painful.  The conversion is less painful for code that's 
branched less.  So maybe in the future a lot of Haskell libs 
will be in the superior darcs2 format.

what an unpleasant situation!  But cross-converting between 
darcs and git format for the same repo is probably even worse.

Last time I tried the darcs-all script (maybe a month ago, 
using darcs 2.0.2), IIRC, it hung, or had some other problem 
in one of the libraries.  Even though it was a clean copy 
that I'd only ever pulled into (many times, and was getted 
by darcs-1.0.9, but still).  And darcs-all on the libraries 
has always been a slow sequential task.  So I'm not actually 
all that enamoured of darcs for ghc development, even for 
the libs.  Since I couldn't update anymore (despite going 
into ghc-head/libraries/something and mucking around with 
darcs-revert and such), I just deleted the tree and decided 
to wait until GHC switches VCS before getting a new copy. 
(trying git-cloning ghc.git sometime, took about 10 minutes, 
nearly no CPU time, and 80 MB, so I'm pretty happy about 
that random experience, but I didn't try to do anything with 
the repo)

-Isaac