Distributed Version Control for ports(7) ( was: Re: autoconf update )

[Ion-Mihai Tetcu]

On Sun, 19 Sep 2010 02:38:28 -0400
jhell <jhell at DataIX.net> wrote:

> On 09/18/2010 07:17, Ion-Mihai Tetcu wrote:
> > 
> > I'm still to see a concise, clear, precise, listing of advantages
> > that switching from CVS would bring us,
> > that would overcome the effort needed to do it (committers, users,
> > infrastructure, tools).
> 
> 
> 1). http://bit.ly/d5UrtN
> 
> 2). http://www.keltia.net/BSDCan/paper.pdf
> 
> 3). http://bit.ly/97
> Y8Xi
> 
> Make your final comparison here:
> http://bit.ly/cyQBn8

>>
concise, clear, precise, listing of advantages, that switching from CVS
would bring _us_
>>

I have to work daily with 3-4 (D)VCSes for my work and OSS work, so I'm
pretty well aware of some good and some bad points of each.

> 4). Because CVS just does not do any of this.

Neither does any of them make coffee or pick up girls for me, but this
neither here nor there, since we're talking about advantages - of
switching - for ports.
General "this is why $VCS is the coolest" and general features matrix
are only the starting point.

> For the sake of argument can you think of any reason to not switch ?
> lets hear those, I'm interested.

Well, first of all, since we are already using CVS, anyone wishing for
a change will have to do the work to break the status quo (ie. convince
the rest that is worth the effort).

Quick args against:
1. Human side:
- all existing committers know to use CVS
- we have a few people that know its internals very well
- most of the user base also
- CVS is simple to use (yes, simple that any of the other; partially
  because it lacks "complicated" "features")
2. Infrastructure:
- everything we have revolves around CVS, from pointy to tindy to
  portsnap to mirrors to ...
3. All the scripts / apps / ... out there that make use of it or csup.

About the only two things I see that we could benefit from switching to
something else are:
- easy move/rename while preserving history (repocopies now)
- better speed for a whole tree checkout/update (if)

I've watched the src switch from CVS to SVN, and I can't say it was
fully a success. part of the problem is that even after all this time,
people haven't completely made the switch in their minds.
And the switch implies much more that a table of command equivalencies.

Anyone wishing to push for a change will have to:
1. Produce a list of shortcomings of CVS in relation to our ports.
2. Produce a comparison of other VCSes in relation to CVS, CVS'
shortcomings in relation to our ports, and each other.
3. Import the existing ports CVS history in the VCS they'd recommend to
switch to.
4. Produce a tested migration path (exporter to CVS that works, etc.).
5. Produce a tested migration path for part of the pieces in our
infrastructure.
6. Document 4. and 5. and CVS to $VCS user giude and be available to
run/fix things related to 4. and 5. for months if not years.

1. to 4. are prerequisites of any serious endeavour to convince our
committers and users (and pormgr@, core@). This implies a few month of
work, without any guarantee that it won't be for nothing.

-- 
IOnut - Un^d^dregistered ;) FreeBSD "user"
  "Intellectual Property" is   nowhere near as valuable   as "Intellect"
FreeBSD committer -> itetcu at FreeBSD.org, PGP Key ID 057E9F8B493A297B
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-ports/attachments/20100919/6d012fb0/signature.pgp