[Mostly to Charles Wilson] Upgrading Cygwin's CVS

[Charles Wilson]

Max Bowsher wrote:
> In http://sources.redhat.com/ml/cygwin/2002-06/msg00754.html, Charles
> withdrew a test cvs-1.11.2 package, saying that some bugs had been found.
> I've recently compiled cvs 1.11.4 for myself, because I wanted the new rlog
> command. I was wondering what these bugs were, in case I might encounter
> them in my locally compiled version.

Gosh, I don't remember the exact details, and can't seem to find it in 
my TODO or NOTES files for cvs.  Trolling thru my mail archives...

It seems that the problems were the standard text/binary issues, on 
reading .cvsignore, .cvsrc, .cvspass -- coupled with issues reading the 
ENTRIES, REPOSITORY, and ROOT files in the CVS dirs.  Something like 
they tended to gain more and more ^M's at the end of each line...which 
led to problems.  These issues are NOT problems on cvs-1.11.0, IIRC, and 
represent a regression for cvs-1.11.2.  Plus, there are the continuing 
problems of hosting a cvs repository on a text mount.  I think.

Really, these issues are not too difficult to track down and fix, but I 
decided to abandon the official cvs codebase at that point(see below), 
and haven't worked up the gumption to re-do all of the original 
cygwin-porting stuff with regards to the cvsnt codebase, so we're still 
languishing at cvs-1.11.0.

Anyway, I'm stunned to hear that the bozos running the cvs project 
actually got around to releasing TWO new versions (1.11.3 and 1.11.4). 
[No, I don't have a lot of respect for people who contemptuously ignore 
patches without even the courtesy of a response...after a couple of 
reminders over several weeks...]

Because of all that, I'd pretty much decided that the next time I update 
the 'cvs' package, I'm going to use the cvsnt codebase (which, despite 
its name, does compile under unix: on "unixoid" platforms, it is 
essentially regular cvs + bugfixes.  Bugfixes the "real" cvs maintainers 
seem to believe are beneath their dignity.  No, I'm not bitter.)  But 
that's a lot of testing I'm not really ready for right now.

So, in answer to your question, I'd make sure that the behavior and 
contents of the .cvs* files, and the CVS/* files, make "sense" when your 
home directory and working directories are on both binary and text 
mounts -- and continue to make sense after a few rounds of commits and 
checkouts.

--Chuck




--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/