The Hermit Hacker <scrappy@hub.org> writes:
> My understanding (and the way I've always done it) is that:
> cvs checkout -P pgsql
> Will remove any old files, update any existing, and bring in any
> new...always worked for me...
Hmm. Now that I read the cvs manual carefully, it does say:
: Running `checkout' on a directory that was already built by a prior
: `checkout' is also permitted, and has the same effect as specifying the
: `-d' option to the `update' command, that is, any new directories that
: have been created in the repository will appear in your work area.
But the more usual way to do it is with "update". Maybe the "checkout"
method is broken, or has some peculiar requirements about how to
specify the repository --- ordinarily you don't need to name the
repository in an update command, since cvs finds it in CVS/Root.
I don't know whether the same is true for the "checkout" method.
Could there be some kind of conflict between what you said on the
command line and what was in CVS/Root?
> Geez, sounds like someone with enough knowledge to build a
> 'AnonCVS Instructions' web page? :)
Actually I'm just a novice with cvs. OTOH a novice might be just the
right person to make such a page ;-). Let me see if I can gin up some
text.
Do you want to adopt the Mozilla method where there is a tarball
available with the CVS control files already in place? I'd recommend
it; my initial checkout over a 28.8K modem took well over two hours.
Admittedly I forgot to supply -z, but judging from the pauses in
data transfer, overload of the hub.org server was also a factor ...
-z would've made that worse.
regards, tom lane