Re: Subversion migration schedule (cut-off Fri 18 Mar)

[Daniel Veillard <veillard redhat com>]

On Sat, Feb 11, 2006 at 06:53:25AM -0500, Owen Taylor wrote:
> On Sat, 2006-02-11 at 08:07 -0500, Daniel Veillard wrote:
> > On Sat, Feb 11, 2006 at 11:22:47PM +1100, Jeff Waugh wrote:
> > > <quote who="Davyd Madeley">
> > > 
> > > > With that in mind, and since I couldn't find it on the wiki. I was
> > > > wondering what reasons we have chosen to go with Subversion over other
> > > > popular contenders.
> > > 
> > > Sucks less than CVS, won't take a lot of relearning, uses very similar model
> > > and infrastructure, relatively easy migration, doesn't change our ability to
> > > use other systems alongside it.
> > 
> >   I don't see any doc explaining that. 
> > 
> > > It's an easy, incremental shift that helps us out in the near term, without
> > > getting in the way of testing/using other (more interesting) systems.
> > 
> >   Easy, good, now how ?
> 
> I know the point of your mail isn't actually to ask these questions,

  What ? It *is* the point. I have a few direct contributors to my projects
I know they are not monitoring gnome-hackers. I will have to help them
make the switch either to subversion or to another place. And the answers
are hence crutially important to me. I don't get patches from the people
who will answer me on this thread, but I need to preserve my pool of
contributors to my projects.

> but I'll answer some of them anyways.

  
> > For example I have scripts building cvs checkout tarball automated on
> > crontab, how am I gonna do this with the new software without possible
> > authentication ?
> 
> Non-anonymous authentication is exactly the same as before - SSH.
> 
> I don't know Ross's exact plans for anonymous SVN, but it is only a 
> few minutes to set up anonymous access to a SVN repository using
> mod_svn; it's really trivial to make things available as, say:
> 
>  svn checkout http://anonsvn.gnome.org/svn/libxml2 libxml2

  So will it be done ? can then a cron job update a directory, if yes that
solves one of my questions.

> >  How are people using the Windows platform be able to
> > use it ? 
> 
> Windows support for SVN is quite good. You can run the command line
> tools, or you can use:
 
   So there is a .exe I can point the users to, that's a good first point.

>  TortoiseSVN: Windows shell integration for Subversion. Quite slick,
>    works great. 
> 
>  AnkhSVN: Visual Studio integration for Subversion. (Don't know if
>    your users use Visual Studio.) Faintly buggy, not quite as slick
>    as TortoiseSVN, but very much usable and the close IDE integration
>    is nice to have.
> 
> There's also Eclipse integration for Linux and for Windows.
> 
> There are a few tricks to getting TortoiseSVN and AnkhSVN set up
> to integrate nicely with 'putty' for SSH access; those would be
> useful to have documented somewhere.

  Can the command line tool on Windows use ssh authentication ?

> > Same for MacOS ? 
> 
> No real experience here, but using SVN from the shell on OS X
> should be just about the same as from the shell on Linux.

  sounds good.

> > Will we still have the equivalent of anonymous
> > checkouts ? 
> 
> Above.

  So I don't have a final answer yet.

> > What is the equivalent operation of the usual CVS commands that
> > the people use now ?
> 
> There are quite a few documents on the web explaining this.
> (At the most basic level, cvs checkout => svn checkout. 
> cvs update => svn update, but not everything is *quite* that
> straightforward.)

  okay I will have to google.

> > At this point I'm just wondering how many of my contributors on my project
> > I'm gonna loose instantly, and I don't find this fun at all whatever the
> > problems some of you may have w.r.t. CVS, it's hard for me to feel happy
> > about this plan.
> 
> There's pain in any transition, but CVS => Subversion is about as
> low pain as it gets, especially for people who are "consuming" a
> project, rather than setting up a new project, figuring out vendor
> branches or whatever.

  Devil is always in the details for such a transition, and converting
people brain is always the slowest. Announcing it without doc published is
nonsense.
  
> I honestly don't see the distributed version control marketplace as
> being well enough organized or mature enough to be worth investigating
> yet for a project as large and complex as GNOME, and subversion does 
> offer substantial incremental benefits. Not huge benefits, but very
> noticeable. So, since Ross has the momentum for doing the transition
> to SVN, I don't see any reason to get in the way of that.
> 
> A month or so should be plenty of lead time to prepare documentation
> to make the transition smooth (and *evening* should be plenty of 
> time to prepare documentation to make the transition smooth.) We
> just need to make sure that happens early on.

  If an evening is plenty enough why not have done it *before* the 
cutoff announcement ???

Daniel

-- 
Daniel Veillard      | Red Hat http://redhat.com/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/
_______________________________________________
gnome-hackers mailing list
gnome-hackers gnome org
http://mail.gnome.org/mailman/listinfo/gnome-hackers