[M3devel] m3 CVS?

[hendrik at topoi.pooq.com]

On Wed, Mar 25, 2009 at 09:44:11PM +0000, Jay wrote:
> 
> /Apparently/ cvs to svn conversion is pretty easy and good.

conversion from cvs to monotone works, too.  It's converting
back from monotone to cvs that is apparently a problem.  But I
still haven't heard the current status of the cvssync branch
of monotone development.  That's the branch where monotone
allegedly worked both ways with CVS.

>  
> svn is not "distributed", but at least it automatically keeps
> originals locally so you can diff without accessing the server.

Being distributed, monotone keeps originals locally, too.  And
monotone will diff them for you.

>  
> And browsing history where multiple files are commited
> in one change works, vs. cvs.

As does monotone, and many current VCSs.
>  
> svn branching doesn't really work, but that's probaby ok.

monotone branching does work.

> Maybe they have fixed it, but for a long time, svn
> would not know what changes had been moved to which branches,
> a pretty basic requirement. Perforce does this perfectly.
> Perforce might be a good option if being open source lets
> us get it for free. Otherwise it is expensive. It is an
> excellent product.
>  
> svn is at least somewhat better than cvs.

In http://gravityboy.livejournal.com/39755.html, gravityboy says

: I'm really tired of people pushing the adoption of svn because it's a 
: better CVS. The only people who that argument hasn't swayed are those 
: who don't care what vcs they use, and that's because they don't 
: care about the argument. Despite its name, svn is the status quo now, 
: so do try to compare it to systems that it hasn't already made obsolete.

Apparently he'd like to have svn compared to newer VCSs instead of 
obsolete ones.

In case you don't know, gravityboy is one of the X developers that got X 
doing reasonable autodetection of hardware.  I'm in awe of what that 
group accomplished.

The thing I really like about monotone is that the developers are 
absolutely paranoid about data loss.  It refuses to sync with broken 
repositories, and spends extra time on internal consistency checks on 
the data base during sync.  This provides some protection against 
corruption due to bits dropped on hard disk and such.

They have a rigorous and pragmatic attitude about correctness that I see 
in the Modula 3 design, too.  Except that they developed monotone using C++.

-- hendrik

>  
>  - Jay