On 3/17/07, Sean Coates <[EMAIL PROTECTED]> wrote:

What files are you looking for? If they are actually _gone_ from CVS,
then someone meddled with the CVS server. There are very few people who
have access to the repository, directly. If someone did this, I'll
complain to the systems people, and it won't happen again (and if you
can pinpoint a time when it happened, I'll check the logs on y1 and find
out who did it).
That said, I suspect that nobody deleted anything from the rcs tree.

Yep. I've already been told that the files migrated to
phpbook/phpbook-xsl/ directory.

As for few people who can access repository directly - two years ago
anyone with CVS access could issue admin commands, such as rollback.
You may try yourself unless [EMAIL PROTECTED] fixed that already. If not -
then it is just a matter of time when anyone does and considering the
weak security of CVS over Internet you do not even need to have your
own CVS account - just find a route to a person who does (from IRC,
for example).


> The problem with CVS is that those who was capable to setup it are fed
> up with the process and do not want to dip into another VCS and those
> who failed are just not here. If not the precious time I could argue
> on the topic endlessly, so I better leave a proposal to discuss.
> http://doc.php.net/rfc/rfc-proposal-show.php?id=8

Um no. You're misinterpreting the situation.
Let me state for the record that I personally vastly prefer SVN over
CVS. I've moved all of my personal projects to svn (a long time ago),
and over the next few weeks will be moving my professional (team-driven)
projects—I finally managed to convince the rest of the team.

Now that that's out of the way, let me clarify the situation with PHP CVS.

There are hundreds (thousands?) of people (and dozens of scripts) that
access php.net CVS. Last time the CVS->SVN migration issue was brought
up, I remember Rasmus shooting it down because of this very fact, and
not because of a technical issue. It would be too much work to move
thousands of people from one SCM to another.

We do not need whole cvs.php.net - only PHPDOC subdirectory. Are you
able to gather statistics (or do you have statistics) how many scripts
are accessing phpdoc/ CVS on a regular basis and what are the files
they request?

As for thousands of people - most of them will appreciate the move.
The problem with only few with write access that do not want it,
because they are busy.

If we need CVS copy to read from for thousands and scripts - it can be
easily done with Tailor script I mentioned - make CVS r/o for everyone
but the script, setup Tailor to migrate changesets from SVN to CVS and
enjoy.


Yes, large projects have already made the change, but I don't think
_anyone_ with the appropriate karma at php.net has the time and energy
to see this through. It's very unlikely to happen in the foreseeable future.

All of this to say that I think it's a tremendously bad idea to port
_part_ of the php.net source code base. CVS may have its problems, but
it's worked for the past ~10 years for us.

Wait. You say that because everyone with required access rights are
busy it is a bad idea to port phpdoc to SVN. That's a big problem and
not an argument for me. That it worked during last 10 years is also
not an argument. New tools are to make life easier.


BTW, if someone did screw with the files on the CVS server, there's
nothing stopping them from doing the exact same thing to a SVN repository.

CVS screwups are the reason SVN was born.


So, to resume - what are the reasons against SVN left?

--
--anatoly t.

Reply via email to