On 11/10/11 10:40 PM, Dave McGuire wrote:
On 10/11/2011 09:32 PM, Toby Thain wrote:
> Further, that low-level "mechanism, not
policy" architecture of X is
> one
> of its big points of existence. An X display can look and act like
> ANYTHING.
And that's the problem. For most uses, the detailed design of widgets
doesn't matter. You just want a CHECKBOX here, a TEXT field here, and a
SUBMIT button there. Good enough for Amazon, at least, which is a
non-trivial UI. Plenty of other examples too.
Of course. But that's a website. We're talking about windowing systems
and local (or "slightly remote") apps.
What's the difference between a web site (which blurs readily these days
into "web application"... Google Docs... GitHub... Twitter...) and a
"remote app"?
You are picking nits, because you are a troll. YOU know the difference
(at least I sure hope you do!), and you know that I know the difference
as well. Again you imply that I think the web sucks, X11 is great, and
X11 should replace the web, and that the web is actually replacing X11.
If this is in fact what you believe, you are incorrect on all counts.
X11 as a remote application server is almost gone, that's just the way
things went.
Further,
toolkits that sit ON TOP OF X give you all that functionality.
They starte with Xt, went through Motif, going all the way up nowadays
to gtk. CHECKBOX here, TEXT field there, and a SUBMIT button
there...except that it's actually fast, and looks a hell of a lot
better.
Then why doesn't the web look just like Motif, since the web offers
everyone unprecedented aesthetic freedom??
Again with the trolling. Nothing looks like Motif, but a whole lot of
web apps look just like GTK.
If the freedom is there, and Motif looks better, why doesn't it look
like Motif? Honest question, not trolling. I don't have time to take
that up as a new hobby.
Until the www shows itself too embarrassingly
incapable of this, you
are
going to be stuck with "cloud" app providers like Google.
*I* am? No, I'm not. I use OpenOffice (erm, "LibreOffice", thank you
Oracle) for stuff like that. Anyone who embraces the idea of storing
business-proprietary documents on someone else's network, especially
when they happen to be experts in large-scale data mining, needs to have
their head examined.
I'm not saying it's clever, or that I do it myself. But it is happening
on a massive scale.
People are moving to web-based word processors, spreadsheets, and
vertical-market software, replacing real software, on a massive scale?
Where, exactly? Prove it. Provide company names and number of seats.
"Test" installations and Utopian "future office" deployments
don't count.
Yes, SMEs are. It began just over a decade ago.
Can't? Didn't think so. Ok, provide OS names and release numbers. This
would need to be OSs that people actually RUN, so Chrome doesn't count.
Um. Web browsers run on more than Chrome. Even Microsoft admitted the
web was important ... 15 years ago.
(Lisp doesn't have any performance problem, but
that's another
thread...)
You haven't done much Lisp, have you. ;)
Quite enough to know you're wrong on that. :)
Ok, I'll trust you on that.
And have you seen what they've done with
JavaScript performance lately?
And noticed that all these supposedly exotic languages are often faster,
now, than old, boring C? Look around... Lisp family languages included.
Seen what Erlang can do? And, more importantly, they're much more
productive to develop in.
Interpreted JavaScript has never been faster than C on the same
hardware. If you disagree, provide proof. JIT-compiled JavaScript has
the potential for being NEARLY as fast, but I have my doubts. It's an
object-oriented language running on hardware that is inherently
procedural in nature. Bridging that gap involves overhead.
There are several interesting cases where very high level languages
outperform C. But that's only one metric.
The world is full of people who claim that their favorite language is
faster than everything else. Most of them are wrong.
That's not the point I'm making. It's also a very uninteresting contest.
And for the record, my favorite language, which is slower than just
about anything else but that's not why it's my favorite, is Scheme. I
don't develop in it professionally because it's...well, too damn slow!
We have something in common then; I like Scheme a lot. But to say it's
slow is silly. Many Schemes compile to C.
There were certainly windowing systems that had some
technical
advantages over X11, but X11 won. It's ironic that the big negative
about X11 was always its bandwidth requirements, but web-based UIs'
bandwidth requirements are orders of magnitude worse, and nobody notices
this and complains about it because people with that much technical clue
nowadays are few and far between.
Do you expect to be using X11 in *another* 20 years?
I have no idea. But you can damn well bet that I'll be a part of guiding
where it goes in the intervening time, because I'm not spending my days
floating along with whatever trendy, bloated, buzzword-of-the-week thing
the trenchcoat-wearing, clove-smoking kids are programming...erm, I mean
"writing scripts" in this week.
One thing you really need to get through your head is that I'm not
suggesting, nor did I ever suggest, nor WOULD I ever suggest, that X11
should replace web browsers.
No, I didn't think you said that. But you did seem to think the web was
an unpleasant step down from X11.
You asserted that X11 is somehow on its way
out (which it isn't) because everything is "moving to the web" (or some
such) now. That makes about as much sense as suggesting that nuts and
bolts are on their way out, because the world is moving to corn on the cob.
X11 *as a remote computing service* is almost dead and buried. X11 in
general is on its way out, but quite slowly. Not that I care much, I
have spent the last decade mostly on more modern systems.
This discussion began with Display PostScript and NeWS, which showed X11
the door, 25 years ago.
JSON is a compact format but not the only one
possible.
I wouldn't call it compact at all, but that's a relative matter. I deal
with lots of binary protocols.
There's a lot of representations that binary isn't terribly good for. A
good use for binary is, as you mentioned, gzipping a text format. :)
So make it all fat and inefficient, and let a compression algorithm
brute-force it back into smallness again. If you were working for me,
I'd fire you.
Eh, okay, but you may have missed my point.
Before you
object - There are exceptions, of course, where binary is a
good idea...
This is the first sensible thing you've said in a week, with the small
modification of the exceptions being the rule. Computers are binary
things. Some computers, notably IBM mainframes and PDP-11s with the CIS
option installed, can operate on text efficiently. Most cannot, and we
pay for this overhead with poor performance and needing multi-GHz
machines to do the very simplest of things.
The PDP-11 popularised byte addressing, indeed. I own quite a few.
--T
That's a very Microsoft-like attitude to be coming from you. I guess
you're not as anti-Microsoft as your rants would have us believe.
I have had enough of your trolling and your feigned cluelessness. I
believe that you actually do know quite a bit about this technology, but
you keep making unbelievably stupid assertions in an apparent effort to
stir up a fight. If you were here, I'd punch you in the mouth repeatedly
and send you home crying to your mommy. You're not, though, so I'll just
tell you to go fuck yourself until you want to talk about classic
computers.
-Dave