RE: Cygnus Win32 B17.1: Port of Cyclic Software CVS 1.9

["Griswold, Victor" <VGris at aironet dot com>]

Mr. Kingdon:

----------
From: 	Jim Kingdon
Sent: 	Friday, February 28, 1997 6:41 PM
To: 	Griswold, Victor
Cc: 	gnu-win32@cygnus.com; bug-cvs@prep.ai.mit.edu
Subject: 	Re: Cygnus Win32 B17.1:  Port of Cyclic Software CVS 1.9

First of all, let me thank you for being willing to work on CVS, and
for the effort that you have put into things like being willing to 
run
the "CVS sanity" tests.  Even if the patches are not yet ready for
inclusion in the main CVS source base, they can still be distributed
elsewhere.  That doesn't imply that the patches are "wrong" or "bad"
or that your efforts are unappreciated.  I do think that making it
possible for CVS to build with Cygwin is A Good Thing (TM).

You are welcome.  Being able to run through the validation suite was a 
requirement from management here in order to demonstrate a lessening 
of risk in our migration from PVCS to CVS/RCS.

As for your comments about Cyclic, if this is going to continue to be
a matter of arguments and flames, I'd rather stop now.  But perhaps I
can point you in the direction of some resources that you perhaps 
were
not aware of, and clarify a few issues:

I have, and never have had, any intention of continuing arguments and 
flames.  I stated clearly that I am willing to spend more time with 
the CVS maintainers communicating about these bugs/fixes.  However, I 
still believe that a couple of hours of 'diff' reading would be 
extremely cost-effective for the purposes of stabilizing the main 
distribution of CVS (Sun platform, etc.) so that it would pass "CVS 
sanity" and correctly handle wrappers of binary files (not thoroughly 
tested by sanity.sh; we stress-tested it under _much_ more demanding 
conditions).  For example, I believe you will find a potential 
resolution to the "can't import binary files under Windows-NT" bug 
listed prominently in the CVS 1997/03/03 BUGS file (and, if you don't 
mind a rather slow turnaround time at this point -- see below, I can 
work with you on incorporating the fixes in a "by the book" process).

> Your email address indicates you work at Cyclic Software, supposedly 
the
> maintainers of CVS.

Cyclic are not the only maintainers of CVS.  CVS is maintained by a
group of people only some of which work at Cyclic.  I know that some
of them read bug-cvs.  If someone else (even someone else at Cyclic,
although I am the only full-time person at the moment), wants to look
at the patches in question, I say *GREAT*!  It wouldn't even have to
be one of the maintainers, if someone wanted to take the patches and
work on packaging them up as described in the HACKING file.

Yes, I did read the HACKING file.  My original intention was to try to 
package each-and-every bug fix separately as requested in the HACKING 
file.  However, due to the rather significant number of bugs (no 
offense meant) and the amount of time spent forcing NT to run 
sanity.sh (no fault of CVS; it is part of the NT file system drivers, 
as commented about in my original posting), I am behind schedule on my 
"real work" and no longer authorized to spend any more bulk, 
all-at-once, time on CVS.  I can, however, spend some time working to 
answer questions about and partition the diffs on an _as_needed_ basis 
(as stated in my earlier email).  In other words, if you (or one of 
the other maintainers) wish to prioritize your interest in the 
non-WinNT-specific bugs, let me know and I can attempt to trickle 
through them over the next few weeks.  Of course, by that time, you 
could... :-)  (this offer of continued work is real, not sarcasm)

> Perhaps it is because Cyclic Software has already found out about 
and
> fixed those bugs internally

We don't have any internal code base distinct from what you see in 
the
nightly snapshots on ftp.cyclic.com.  In addition to the snapshots,
you might be interested in the commit-cvs mailing list (see HACKING
from a snapshot of newer than 4 Feb 97).

It is good for everyone that you do not have private branches for each 
major customer; that benefits all users and is a practice to be 
commended.

I was unaware of this mailing list, since my snapshot was from some 
time in mid January, after 1997/01/15.

> Perhaps it is because with a reliable piece of software, there will 
be
> less of an incentive for people to purchase a maintenance contract
> from Cyclic

I don't know of any way to prove or disprove that notion, but I will
offer my opinion: if we did that, then we might lock ourselves into a
large share of a tiny market, but we would insure that CVS remains a
bit player.  I'd rather have a smaller slice of a large and growing
CVS pie.

That is good news.  I must say that there has been some concern 
expressed in this mailing list about "free software" gradually 
becoming less and less free (only Cygnus' future track record about 
timely releases of Free cygwin32, under GPL, will demonstrate by 
action what their intentions are:  and I personally suspect that 
things will be OK).  With respect to slice-of-the-pie, note my comment 
above about migrating our development efforts away from PVCS and 
towards CVS/RCS.

I also assure you that the reasoning behind requirements like needing
to separate unrelated patches into separate submissions (as described
in HACKING) _IS_ the desire to make CVS as reliable as possible.

I can understand this as the ideal case, but, when one has many bugs 
at once in a product one must expect that, at times, there will be 
many bug fixes encountered at once (and a tested bug fix is generally 
much better than a mere cry of "it's broken!").

> Perhaps some of the bugs have been fixed through contributions on a
> mailing list, but I have been unable to find an archive of the CVS
> mailing list.

There are some archives of info-cvs at
http://tongue1.itc.nrcs.usda.gov/ (the page calls it "cvs-info"). 
 I'm
not sure I've seen any bug-cvs archives.  Perhaps some volunteer 
would
be interested in archiving bug-cvs and making that archive available.

Thank you for pointing me to these archives.  This location would be 
good to put in the CVS 'README file.

I hope you understand that Cyclic is a very small company with very
limited resources (see the annual report on our web site for 
details),
which necessarily must be focused most strongly on what our customers
are asking us for.  We do try to do things which benefit 
non-customers
as feasible, but we can only do so much.  We have tried very hard to,
as much as possible, set up CVS maintenance so that this does not
result in us being a bottleneck.

> Ah, [NT tkCVS] port just showed up on 2/17/97 (since I was last on
> the Cyclic site).  Thank you for the information.

Hope you find it useful.  I don't know much about it other than what
is on the web site.


Victor J. Griswold, D.Sc.
Aironet Wireless Communications, Inc.
voice:	330-664-7987
fax:	330-664-7301
email:	(MS-Mail) vgris@aironet.com
	(MIME) Victor.Griswold@pobox.com

application/ms-tnef