Re: CVS commit: pkgsrc/net/ocamlnet

[David Holland <dholland-tech%netbsd.org@localhost>]

On Sat, Jul 21, 2012 at 05:30:53PM -0400, Gary Duzan wrote:
 > In Message 
 > <a0806cf39566eab5ed139f0478fb5555.squirrel%www.duzan.org@localhost>,
 >    gary@duzan.orgwrote:
 > 
 > =>=> On Thu, Jul 19, 2012 at 01:17:35PM -0400, gary%duzan.org@localhost 
 > wrote:
 > =>=>
 > =>=>  > I've even managed some success doing incremental
 > =>=>  > conversions from CVS (though the CVS repo could use a bit of cleanup 
 > to
 > =>=>  > make it seamless).
 > =>=>
 > =>=> if you mean the NetBSD CVS repository please elaborate...
 > =>
 > =>   Sorry, no. This was a CVS repo we have at work. The main issue there
 > =>was with some files that were removed then readded, with the deleted
 > =>version having a newer timestamp in CVS. The versions after the restore
 > =>were not preserved. Other than that, though, doing incremental
 > =>conversions has been smooth.
 > =>
 > =>   I have attempted to do NetBSD repo conversions in the past, but the
 > =>main issue was memory, as the changeset building code is a pig. I'll
 > =>make another attempt when I get a chance. Maybe it would be a good
 > =>excuse to buy that new hardware with 32GB RAM I've been looking at...
 > 
 >    I made the attempt with my 12GB of RAM against an rsync clone
 > of the CVS src repo, and it made it a fair ways, if slowly, until
 > it hit:
 > 
 > 371622 avoid losing precision warnings
 > 371621 add explicit casts to avoid parameter conversion warnings.
 > 371620 fix non-portable bit-fields
 > 371619 const poisoning.
 > 371618 - Process all passed option rather than ignoring silently except the 
 > first one.
 > abort: malformed response from CVS
 > 
 >    The "const poisoning" version was from July 2000, and seems to
 > have been imported properly. I'll see if I can figure out why
 > exactly it died, and also see how it does with xsrc, pkgsrc, and
 > pkgsrc-wip.

The "process all passed option" changeset is version 1.7 of
src/usr.sbin/wiconfig/wiconfig.c.

The changeset after that (which is potentially the one it aborted on)
is version 1.14 of src/sys/arch/i386/stand/dosboot/Makefile.

I don't see anything particularly unusual about either of them with
either cvs log or cvs diff...

Thanks for trying this.

-- 
David A. Holland
dholland%netbsd.org@localhost