Path: utzoo!attcan!uunet!clyde.concordia.ca!news-server.csri.toronto.edu!smoke.cs.toronto.edu!neat.cs.toronto.edu!moraes
From: moraes@cs.toronto.edu (Mark Moraes)
Newsgroups: comp.windows.x
Subject: Re: Xw on Sparc? Xaw vs Xw vs XView? Interp layout contraints?
Message-ID: <90Mar16.165337est.242@smoke.cs.toronto.edu>
Date: 16 Mar 90 21:53:56 GMT
References: <6873.25ff8b27@swift.cs.tcd.ie>
Lines: 90

jorice@swift.cs.tcd.ie writes:
>(1) Why no discussion of Xw? Is noone using it because it's unsupported, or
>because it's horrible in some way that I haven't found yet, or are all the Xw
>users so serenely happy with their system that they have no need to ask
>questions?

Mostly because it's unsupported, I guess.  I found Xw very nicely
designed, fairly solid, well documented and easy to use.  It got used
in a few programs here.  I still prefer Xw.

There's also the fact that the Xw with R3 didn't work with R3 Xt
without major mods (I'm told Xw sort of evolved into Motif, which could
explain why Xw isn't supported any longer.  Or maybe HP just got tired
of giving away their work) It's unfortunate - many of us hoped it
would become the standard widget set when it came out; technically it
is really nice.

For those people writing freely redistributable programs for X, who
don't care too much about look/feel, Xaw remains the preferred choice
because the people at MIT do fix bugs, it's sort of guaranteed to be
in most X distributions despite the fact that it isn't a Consortium
standard.

I wish the X Consortium would adopt Xw, or at least some of the Xw
widgets to fill holes in the Xaw line -- at present, Xaw is sort of ok
for simple user interfaces; feels kind of clunky.  Xw feels smoother,
and had good documents, a nice class hierarchy.

Motif is the official party line, it would appear.  People who've used
Motif tell me it's very solid, much better than Xaw and Xw, and worth
paying for.  Said people have lots of disk space and memory, of
course.  Doug Young's book is now out in a Motif version.

>(3) After getting the version of Xw from the contrib section of the X11R4
>release tape and compiling it, I was disappointed to see that it's not the 3-D
>version that's used in Young's book (3-D as in buttons looking slightly raised,
>text input boxes slightly sunken, etc.). Can this be got?

Not easily, no.  If you happen to have a pre-R3 copy around (the one
that was on expo.lcs.mit.edu for a while before the R3 release), that
had the 3d look.  It got taken out possibly because of the look and
feel litigation, or perhaps because HP just wanted to keep the spiffy
looks for their product release only.  Since I personally don't see
why 3d is so desirable, I didn't care much.  It's probably feasible to
put it back in (I remember taking the 3d stuff out of the button code
to try and get xpic to start in finite time -- on the R3 Xsun, drawing
the 3d stuff was agonizingly slow)

>(4) On the subject of XView, how is it to program? I'm wary of it because I
>found programming in SunView needlessly difficult and I think that SunView
>applications look really clutzy. How much better is XView? How much disk space
>does it take up when compiled, roughly?

XView seems considerably better than SunView in the programming
interface -- it's got some of the Xt object-oriented approach.  But
it's big.  Some of the code in the internals of the notifier looks as
if it would be painful to port to a system without Berkeley style
signals.  (We have one of those:-) Has anyone gotten XView running on
non-Sun machines?

One possibility for a small toolkit would be to use olgx, and bits of
olwm for the spiffy 3D look.  olgx is fairly small.  Building a set of
widgets with it, or retrofitting it onto Xw might be nice.

Given the great response of the XView people to questions on the list,
and to bug reports, it's might be a good alternative for those of us
too broke to buy widget sets, or those of us that want source with few
strings attached.  (Obviously, just like Xw, there's the possibility
that at some point in the future, XView src may stop appearing in MIT
X distributions)

>(5) Some of the test programs that came with Xw don't seem to work correctly -
>can someone tell me why? Generally, the little sash sub-widgets in VPaned
>widgets don't work right. They always come up as blank white instead of black,
>for a start, but the real hassle is that unless the server is switched to bug
>compatibility mode, things bomb out when you try to grab the sash with the
>pointer. The other odd problem is in the test program menuTest1, where the
>pop-up menus brought up by clicking on the two lowest buttons just flash up on
>the screen and then vanish again. My machine here is a Sun Sparcstation1
>running SunOS 4.0.3c. Yes, I'm running Release 4 of X with at least some
>patches applied. I've tried compiling the problem programs with both cc and gcc
>with exactly the same effect.

Probably bugs in Xw.  Hammering it through lint or a fussy compiler
(gcc -O -W, or the IBM RT/PC hc) would be a step in the right
direction.  Doing a code walkthrough wouldn't hurt -- there may be
things in the R2 to R3 conversion that got missed, despite the
comprehensive job Marty did.

	Mark.