[comp.windows.news] comments on X11/NeWS so far ??

eho@clarity.Princeton.EDU (Eric Ho) (10/10/89)

For those of you who had been using X11/NeWS beta (or even FCS), I wonder what
are your impressions so far -- mainly in terms of reliability, speed &
bloatness (say from 8 meg Sun-3/50's & 3/75's to 16 meg Sparc boxes).  

Also, I heard that X11/NeWS (aka OpenWindows + XView) will come with SunOS
4.1.  Is that true ?
--

Eric Ho  
Princeton University
eho@confidence.princeton.edu

schwartz@psuvax1.cs.psu.edu (Scott Schwartz) (10/10/89)

In article <EHO.89Oct10003844@cognito.Princeton.EDU> Eric Ho writes:
| For those of you who had been using X11/NeWS beta (or even FCS), I wonder what
| are your impressions so far -- mainly in terms of reliability, speed &
| bloatness (say from 8 meg Sun-3/50's & 3/75's to 16 meg Sparc boxes).  

Speed: not bad.  It seems a little faster than NeWS 1.1, but that may
simply be that I'm running it on 3/60's rather than 3/160's.  This is
a highly subjective impression, though, so I may be way off base.
Startup is pretty slow, in general.  Clipping seems faster (i.e.
partly obscured windows are smoother.) Implementing something like
"undump" might help a lot.  The "spinning earth" demo still kills the
server, even on a Sun4/280.

Bloatness: Sigh.  It is fine on diskless 16M 3/60's.  4M 3/50's don't
stand a chance.  I'm talking 15 minutes to startup and a couple
minutes to handle a mouse event.  Bad locality of reference, I guess.
Running with X only or NeWS only may help; I haven't tried it.

Reliability: Some of the X stuff needs fixing up.  I've had reports of
problems with xhosts from some users.  Some NeWS applications die
mysteriously.  For example, pageview chokes on large documents (may be
running out of /tmp space; have to look into that; better diagnostics
would be nice), and seems to have trouble with the output of some tex
to ps translators.

General: Two people I showed xnews to instantly converted to it (from
mit X).  The problems they run into are outweighed by the advantages
of a merged server.

Needed: preemptive scheduling in the server.  Some server operations
take a reletively long time, during which your mouse becomes
unresponsive.  Users become unhappy.
--
Scott Schwartz		<schwartz@shire.cs.psu.edu>
Now back to our regularly scheduled programming....