[clue-talk] Why X? [long, with provocative questions]

[Tom Witmer]

On Wed, 2005-04-20 at 21:18 -0600, Matt Gushee wrote:
> >      2. The first rule of optimization is to not optimize anything until
> >         you have repeatable, quantifiable numbers that you can use to
> >         see if you've made things better or worse.
> 
> A very sound principle. It seems, though, that like many good principles 
> it gets corrupted in practice. One of the folk-wisdom versions of this 
> notion that I've heard is that "performance doesn't matter any more." 
> Which is probably true for major corporations that can solve problems by 
> throwing hardware at them. Less true, and maybe untrue for--yes, home 
> users--but also small businesses, students, non-profit organizations, 
> anyone on a tight budget.

   Too true -- I've seen this used as an excuse before on projects and
disagree with it as well. However, I brought up the point because of
what you said in your original mail: "I can't help thinking that a
significant part of the problem is the venerable X Window System--in
particular, the fact that it is a client-server system."

   X's client-server nature often gets blamed for its poor performance,
but the numbers haven't agreed with that assertion. There are problems,
but eliminating client-server would likely yield 0% improvement, since
that's not where performance analysis has shown the bottleneck to be.
Which of course leads to the question "Where is the bottleneck then?".
That's what I was trying to partially address in the first email.

   Besides client-server is pretty much inevitable at this point. Even
if you eliminate the network layer (at ~0% performance improvement), how
does OpenOffice share the screen with a 3D game ("Cube"), KMail, and
Firefox? Now add drop-shadows, translucency, window overlap, opaque
window drags, and you're right back to building a client-server
rendering system. Once you do that, adding network support is trivial,
and drops performance by approximately ~0%. Congratulations! You've just
re-invented X! :-)

   That said, X is far from perfect, but is only now beginning to get
fixed. Will the fixes be enough? I wish I knew, but the chatter on the
mail lists seems encouraging. Time will tell.

   
> >      6. Scalable Vector Graphics (SVG) and Compositing (translucency,
> >         drop-shadows, lots of wild eye candy) are on their way as well.
> >         SVG will provide some serious usability enhancements, as well as
> >         allow for even more wild eye candy. (Hey, some of that stuff is
> >         actually useful!)
> 
> Hmm ... what do you hear about SVG and usability? I was an XML 
> consultant for a short while ... they said I was good, though I don't 
> know why ... later became disgusted with the XML world and have been 
> pointedly ignoring it for the last 2 years or so. But I still have some 
> interest in the field, and I always thought SVG was kind of cool.

   Yep, XML is one of those things that lets any technology be
"reinvented" by adding a "with XML!" sticker on the end. I like
simplistic use of XML, but anything complicated tends to be better
served by different (usually pre-existing) technologies.

   I don't know that much about SVG, but the resolution-independent
rendering should allow for rescaling of text and icons without ever
worrying about jaggies again. I think about this whenever I visit the
parents and do the sysadmin thing with their PC. It's set to something
like 640x480 resolution which drives me crazy, but any other setting is
too small for them to comfortably read. Something in-between would be
better, but not possible at the moment.

   SVG might allow for some neat Gimp enhancements as well, but now I'm
well and truly in an area I really know absolutely nothing about! :-)

 - Tom

p.s. In '91, I witnessed the construction of a "FastMath" library that
was 1) shipped 10 weeks late, 2) Inaccurate, and 3) Slower than the
stock math libraries. Ah yes, optimization is the friend of the time-
and-materials contract! These days of fixed-price bids have actually
stopped some of these horrors, although they bring their own set of
different horrors.

-- 
Tom Witmer <tom.witmer at comcast.net>