ndad import failed -- can someone please discuss with me?

Sorry about the lack of response on #launchpad, but many of us are not in the UK time zone.

I'll attempt an import now.

The import failed with:

CRITICAL: requested checkout of reporoot '/var/local/cvs_mirror', file 'NDAD/fix2csv/Attic/README', revision '1.2'
CRITICAL: log of checkout responses follow.
CRITICAL: 'E cvs [checkout aborted]: there is no repository /var/local/cvs_mirror/NDAD/fix2csv/Attic\n'

I'm trying again just in case it is a transient failure.

On Thu, Feb 28, 2008 at 07:00:56AM -0000, Tim Penhey wrote:
> CRITICAL: requested checkout of reporoot '/var/local/cvs_mirror', file 'NDAD/fix2csv/Attic/README', revision '1.2'
> CRITICAL: log of checkout responses follow.
> CRITICAL: 'E cvs [checkout aborted]: there is no repository /var/local/cvs_mirror/NDAD/fix2csv/Attic\n'

Hm. There is no Attic directory under the fix2csv dir in the repo.
There is also no revision 1.2 of fix2csv/README; according to the
repo, the latest is 1.1.1.1.

Any idea what we can do about this?

Ben

You could get rid of the fix2csv subdir completely, or you could try reconstructing the Attic dir and see how that works out for you.

The thing is, I can't reconstruct the Attic dir, there never was an
Attic dir. The fetcher is trying to fetch a version of that fix2csv/README
(1.2) that does not and has never existed -- that's the only reason the
CVS repo is looking for an Attic where there isn't one, it will presumably do
the same if you ask for version 99.9 of any file. The question is
why is it trying to do that? It seems like it might be a bug in the fetcher unless
there's something very weird about my repo, but it generally hasn't been messed
about with by anything other than CVS itself.

I see what you mean. That's odd. It /could/ be a bug in cscvs (you can check out the sourcecode at launchpad.net/launchpad-cscvs) but it could also be something weird in your repo that you're missing (cvs repos are as much fun as the big hotel in the shining!)

Maybe you can do some grepping there to check?

Alternatively, just move fix2csv out of the way and I'll have the import run again. If it works, there's some evidence that there may be something amiss in there.

I did already grep the history file and look at the logs for the file, nothing untoward there. Shrug. I've removed the directory and all reference to it from the history file, so please kick it off again and we'll see what happens...

Kicked off.

And.. imported! Congratulations. You might want to file a bug at https://launchpad.net/launchpad-cscvs/+filebug and attach a tarball of the contents of fix2csv to make sure we keep track of the problem for future reference. PS: I hope you didn't really mean remove when you said "removed"!

Thanks!

What you're mirroring from is itself a mirror of our CVS repo, so yes, I did delete it entirely from there, but not from the original repo. What I'm planning to do is move fix2csv (using 'cvs rm' and 'cvs add', at the cost of losing the old revision history for it but I can live with that) and hopefully the import will still work...

I notice that all of changesets 3190-3373 are labelled "Filler changeset 1.1->1.1.1.1" and don't seem to contain any actual file changes. What are they for?

Ben,

I don't personally know what the Filler changesets are for, but I asked one of the original developers. He seems to recall that it has something to do with either concurrent commits to the CVS tree or files that were created on branches.

The code that is used for the imports can be found at http://bazaar.launchpad.net/~launchpad/launchpad-cscvs/rocketfuel/files.