Date: Tue, 16 Apr 1996 14:44:47 +0100 (MDT) From: Jeremy Chatfield <jdc@crab.xinside.com> To: andreas@knobel.gun.de (Andreas Klemm) Cc: ports@freebsd.org Subject: Re: Xinside's Motif Message-ID: <199604162044.OAA13685@crab.xinside.com> In-Reply-To: <Pine.BSF.3.92.960414120642.226E-100000@knobel.gun.de> from "Andreas Klemm" at Apr 14, 96 12:09:20 pm
next in thread | previous in thread | raw e-mail | index | archive | help
Andreas Klemm writes:
> On Fri, 12 Apr 1996, Jeremy Chatfield wrote:
>
> > [...] On FreeBSD, with a variety of different mallocs,
> > using the same source code, the MWM+VTS memory use totals around 45MB.
> > We think that the problem is therefore external to our Motif, and may
> > lie in the FreeBSD libs, or perhaps is some bizarre effect of the VM
> > system. [...]
>
> Just one question. Will your port be based on FreeBSD-current or
> the latest SNAP using Paul Henning Kamp's new malloc implementation ?
>
> Or do you use a FreeBSD-stable (2.1 + bugfixes) release ?
We used 2.0, 2.0.5, 2.1-RELEASE and, I believe, 2.1-STABLE at various
stages. The compiled version to be passed to Jordan as a release candidate
was done on 2.1-RELEASE, according to the programmer. The Motif
Validaqtion Test suite and some individual use has shown no Motif-related
differences between all of these. Obviously, we'll claim 2.1-RELEASE and
2.1-STABLE as supported, eh?
In general we regard it as unwise to port to a moving target like
'-current'... '-current' as of when... It's pretty hard to support. "Hi,
umm, you were running 2.2-current? OK, uhh, when did you grab that? Oh, I'm
sorry we can't offer support, you grabbed that three days too late." ;-)
Seriously, one of the reasons we like the *BSD's, despite their smaller
market share, is that they are relatively more stable than the Linux
distributions, who constantly shuffle files, installation methods, etc.
It's the only reason we can afford to keep porting and maintaining them all,
really. Please don't start this silly Linux-thing of running Beta libs, new
kernel VM subsystems and then expecting a commercial vendor to have already
grabbed the two-hour old code and run a complete validation test suite on
it, have the release notes and all sales and support staff prepped with the
asnwers about why it does/doesn't work and what the recovery plan is. It's
not funny any more :-(
The malloc is from the dynamic lib - whatever that is. When testing,
we used our Server malloc, the GNU malloc, the older FreeBSD malloc and the
new Poul-Henning Kamp malloc (which bears some similarities to our Server
malloc and the new Linux malloc by H J Lu). I believe that the most recent
testing was done with the P-HK malloc. Not that it made a blind bit of
difference - the bloat that I discussed was pretty much identical with all.
Cheers, JeremyC.
--
Jeremy Chatfield +1(303)298-7478 FAX:+1(303)298-1406 email:jdc@xinside.com
Commercial X Products - for more information please try:
X Inside Inc, 1801 Broadway, 17th Floor, Denver, CO 80202
http://www.xinside.com/ majordomo@xinside.com ftp.xinside.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199604162044.OAA13685>