Tim Endres wrote, quoting me:
> > I develop on Windows. I have compared your jCVS 4.7 and wincvs.org's
> WinCVS
> > 1.0.6 and found jCVS to be vastly inferior, and I'm cross-posting to
> > [EMAIL PROTECTED] and comp.software.config-mgmt. I'm recommending WinCVS
> to
> > everyone. I don't have a legal or ethical problem because both jCVS and
> > WinCVS are distributed under the GNU Public License ("activities other
> than
> > copying, distribution and modification are not covered by this License;
> they
> > are outside its scope").
>
> Sounds fine. I would post a reply stating that people should try 5.1.
> It competes quite well with WinCVS.
>
[Randy Stafford] Ah. So the playing field wasn't exactly level,
because I compared the most recent version of a competing product with an
older version of your product. You'd feel compelled to make that point in
defense of your product's reputation and that of its developer(s). But by
that time, some damage might already have been done, because the tilted
comparison would have been made public, and could potentially influence the
thinking of future technology selectors who might not have the opportunity
or inclination to consider your rebuttal.
This is the position that GemStone finds itself in. The Charles
University EJB server comparison published last week used GemStone/J 3.0.2,
which was outdated four months ago by GemStone/J 3.1 (itself outdated
yesterday by GemStone/J 3.2). Let me quote to you from the opening
paragraphs of the release notes of GemStone/J 3.1: "the principal focus for
the 3.1 release of GemStone/J has been to optimize the product for
performance, scalability, and reliability. We've made a large number of
improvements in all of those areas... Performance has been improved across
the board in GemStone/J 3.1... In general, we have improved internal
designs, reduced unnecessary internal communications, optimized algorithms,
and removed unnecessary code. We expect just about every kind of
application to see noticeable performance improvements from this release,
without any changes to application code." I have personally observed 3X
improvements between 3.0.2 and 3.1 in the application I'm working on - out
of the box. And 3.2 uses the HotSpot VM by default, which further improves
matters. Among other failures, the Charles University comparison fails to
account for the performance improvement effects of these releases.
> Where is the problem, Sir?
>
> Only in your *intentional* spreading of misinformation. Which has nothing
> to do with an honest, third party, details fully disclosed test.
>
[Randy Stafford] I want you to know that I have not posted any CVS
client comparison conclusions on [EMAIL PROTECTED] or
comp.software.config-mgmt, nor do I intend to. I was merely hypothesizing
an analogous situation to illustrate the above point. I can see by the
level of concern evident in your reaction that you care about your product's
(and company's) reputation, and you care that it is fairly evaluated and
publicly critiqued. So do we, in regards to our product. No one would
expect anything less of either of us.
Frankly I have never undertaken a serious evaluation/comparison of
GUI CVS clients. I did try jCVS a little less than two years ago, but
abandoned it for the command line. I didn't have a major problem with it
per se; it just seemed that the GUI was a little clunky and I was more
efficient on the command line - therefore jCVS was useless to me. And I am
currently using WinCVS, because it adds value while still providing a
command-line mode and is therefore useful.
But suppose I did undertake a comparative evaluation of GUI CVS
clients. Suppose I evaluated the candidate clients on a machine with too
little memory to accomodate the Swing-induced footprint increase of jCVS 5.1
over jCVS 4.7.5's AWT footprint (acknowledged in the 5.1 release notes), and
suppose I concluded that 5.1 performed poorly as a result. Suppose some of
my criteria were support for arcane CVS commands that are never really used
in practice and might someday be phased out, and suppose jCVS got low marks
in this area (I don't know that it would). Suppose I was over-emphasizing
tight Windows integration and naively ignoring the broader set of usage
scenarios afforded my be the portability of jCVS. Suppose I published my
conclusions without first giving you an opportunity to point out these
issues, and point out to me how jCVS 5.1 "competes quite well with WinCVS",
as you put it. The result is that damage might be done to your product's
and company's reputation. For better or worse, this is the kind of thing
against which the license clauses to which you object are intended to
protect. The Charles University EJB server comparison contains analogous
shortcomings with respect to its treatment of GemStone/J, as pointed out the
other day by Lougie Anderson, our VP of Engineering.
> Frankly, I will place my lot with my code. Which has proven to be quite
> a successful strategy to date.
>
[Randy Stafford] As will we. GemStone has been shipping object
application servers since 1987, business has never been better, and our
trajectory will continue to sustain and improve our position in the market.
> > PPS. Your domain name implies that you want people to trust you, and
> yet
> > you're openly announcing that you'll litigate the validity of software
> > licenses?
>
> Randy, please, I am feeling bad for you now. I hope that people would
> trust
> me not to lie down and take it from an overzealous ministry, er, I mean
> vendor.
> I guess it depends on your definition of trust. I trusted that report to
> remain
> available on the internet, and now it is gone.
>
[Randy Stafford] Don't bother feeling bad for me - it's wasted
emotion. Obviously you're not the type to lie down, and we're not trying to
make you or anyone else "take it". Nor are we an "overzealous ministry" -
we're just defending the reputation of our product and company, as you did
with your reply to my message. And we're just as proud of the Smalltalk
tradition from which we come as you are of the Unix and open source
tradition from which you come. It's obvious that you are contributing a
great deal to the Java community, as we feel we are with some of our unique
features, and that's a noble thing. We would all do well to remember that
we are part of the same community - would we rather be programming DCOM?
One last thing - you seem to be making an implicit assumption in
this and other posts that we had something to do with removing that report
from the Internet. As far as I know we did not. My best guess is that
Charles University removed the report, perhaps to redress its shortcomings
which should by now be obvious, or perhaps out of responsible legal concern
over potentially violating not only GemStone's but also WebLogic's license
agreement.
Randy Stafford
Senior Architect
GemStone Systems, Inc.
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
