[Gnu-arch-users] Re: cscvs--experimental--1.1 nearing doneness; call for testers

[John Goerzen]

Charles Duffy <address@hidden> writes:

> I'm getting ready to wind down cscvs--experimental--1.1, and would like
> to declare it (hopefully) featureful, reliable and stable enough for a
> wider body of guinea pigs^W^Wusers.

Excellent.  I decided to try to import the Linux kernel tree from
bkcvs using your cscvs.  This ought to be an excellent test case :-)

First the good news.  This is the first time I have had cscvs cache -b
run to completion on *any* branch of cscvs.  The others either ate all
the system RAM, or alternatively used more than 10GB of disk space for
the catalog.

Now the bad news.  cscvs starts by generating at least one -- and
perhaps two -- changesets for the initial add of each file.  That is,
it starts out wanting to commit -- get this -- 30,000 changesets to
tla just to add the files from the initial cvs import.  That's not
acceptable at all.

After that, it identifies approximately 10,000 changesets that look
more reasonable up front but that number still does seem suspiciously
high.

FWIW, the other cscvs trees appeared to exhibit the same behavior wrt
all those initial changesets during their cache -b procedures.

I am using the archive from
:pserver:address@hidden:/home/cvs, module linux-2.4.
Feel free to test locally :-)

-- John