Joe
dcoombs told me a couple of days ago that the server where our svn was
hosted has been hacked (likely via a weak password on ssh or rsync),
so they shut it down to prevent further damage.
I now have a copy of the svn repository and will be trying to put it
up "somewhere" soon - probably today or tomorrow. I figured this
would be a good opportunity to rearrange things and put them up on
their own sites. Particularly schedulator probably deserves to be
separated out on code.google.com or somewhere.
Sorry for the inconvenience.
Have fun,
Avery
I think I heard that such a thing existed. Thanks for the link.
> If we want to rearrange things, we should probably use a dummy
> repository as the destination for some dry runs, before targeting
> code.google.com itself (it doesn't seem to allow for multiple "resets"
> of the repository, you need to send an email to the admins if you need
> to try it again).
Yes, this is easy enough with svnadmin/svndumpfilter.
> One good way would be to (for each target project on code.google.com:
> wvstreams, schedulator, etc) make a local repository, make it look the
> way we want (including the history), then svnsync that over.
>
> Also, beware, the svnsync takes a *really* long time. Thankfully, it's
> a synchronization, you can re-run it if there's an interruption (you
> have to mind the locking that they do, if it stopped in a rough way).
In the case of schedulator, the import probably won't take long since
there really aren't many revisions (and there are no branches). Other
parts of the alumnit tree are obviously another story.
On the other hand, I'm somewhat considering using pure git instead for
most of this stuff. We were originally talking about having a hybrid
svn-git scheme, but since the alumnit repository died anyway
continuing svn is suddenly more work than just dropping it.
Does anyone have an opinion about git vs svn for schedulator
specifically? It's tiny enough that I can set it up right away,
however we choose to do it.
For wvstreams/wvports it's still more work to sort out, because the
repository including wvports is about 800 megs of git (1.8 gigs of
svn) and most of it is crap... but not all of it, and googlecode has a
100 meg limit for svn. So it'll take some filtering to split it all
up properly. But I think git's "grafts" feature is the secret to
doing this properly:
http://article.gmane.org/gmane.comp.version-control.git/66402 . I'm
still trying to sort it out. Sorry for the delay.
Have fun,
Avery
Hi all:
I've gotten the schedulator repository reposted at google code and
repo.or.cz. See http://code.google.com/p/schedulator .
Let me know if you notice anything weird.
Have fun,
Avery