Re: Completely OT: A look into the furture

Hello Fredderic

Fredderic wrote:
One must remember also, though, that we're no longer living in the days
where you only run one program at a time. The ability to share a
library's memory image between several applications does sometimes mean
that by bundling some extra stuff into one "bloated" library, you end
up with a net total saving. The concept of "efficiency" is indeed a
while lot more complex than it once was.

You are of course right with that. Efficiency is a very flexible term
and it all depends, what you try to achieve. And when I program for my
own fun projects, I also use libraries of course, because it's
inefficient always to invent the wheel again. But software often
doesn't really work like that. A good example are the application
bundles used on Mac OS X, where you simply install an application just
by dragging it's archive somewhere onto your HD. Really cool
feature... But those apps can't be sure, that all needed 3rd party
libs are installed on the target system, so they have to come with
them. Because many such apps do so, you end up with many different
versions of the same lib on your system and loose the advantage of
code sharing. The same is valid for other systems such as Linux or
Windows but for different reasons and different mechanisms behind it.
Libraries and code sharing are a really good idea, and one should use
it, *if necessary and useful*. However the effects of version problems
and other pitfalls often weaken the advantage of library usage.

Anyway, this wasn't the point of my complaints initially. I read Joels
article about the future of web based programming and said, that I
agree in most parts, but I really have problems with the statement,
that just adding enough layers of abstraction and indirection to
distributed AJAX-alikes, layers of JavaScript/...anyscript/VMs and so
on is the only way to go. I think, this leads in the end to
inefficient architectures and to the IMHO bad habbit of many
programmers to create and start up a gigantic machinery of libs,
interpreters and software layers, just to do nothing more as to pop up
a "Hello World"-message. You can argue, that computers are fast enough
to handle all that, so what? But aside from ecological implication of
such a CPU-cycle-wasting it is often really inefficient and lacking
elegance to do so.

Fast enough computers and efficiency: I read somewhere the following
argument: my computer today is 100 times faster than the one 20 years
ago, but the time from switching it on until the user interface is
ready for work *increased* to many times of the boot time for the old
cmoputer. An old Amiga (that was, what I head the old days) with a
small HD bootet in a few seconds to the Workbench and was ready to
use, my fast Linux-Box, Mac or smoe Windows computers need *minutes*
to be ready! Okay, they do 10 times more while booting compared to the
Amiga, but also then they should be 10 times faster, instead of
slower...

So what I say is *not* that we should not use libraries or modern
tools. I say, that adding mindlessly layer above layer of abstractions/
libs/interpreters and VMs between machine and user can't be really the
way to go.

Although it does concern me just a little when the decorations on my
application windows are drawn using a theme engine written in Scheme.

Exactly!

Stephan

.

Relevant Pages

  • Re: Why are .Net UIs slower than C++ or classic VB?
    ... whatever names this kind of things have had (The first computers had no OS), ... direct a problem after the first use of OS that did not exist from layers. ... Therefore in my opinion the Net should *not* replace the Win32 API, ... responsive (see Franklin Covey's PlanPlus for Windows XP or ...
    (microsoft.public.dotnet.general)
  • Re: Handling Bugs
    ... Some are still using Windows 98. ... the system and install it on the computers in the different schools. ... I mail out new versions yearly for them to install. ... Let me know a little bit more about the problem of libraries if you would - ...
    (microsoft.public.access.formscoding)
  • Re: Handling Bugs
    ... Some are still using Windows 98. ... the system and install it on the computers in the different schools. ... I mail out new versions yearly for them to install. ... Let me know a little bit more about the problem of libraries if you would - ...
    (microsoft.public.access.formscoding)
  • Re: vlsi design
    ... and many of the Stanford libraries will let any schmuck off the ... More computers, less books. ... Probably in violation of IEEE terms. ...
    (sci.electronics.design)
  • gates la rrruminii salbatici si primitivi
    ... Romania ... libraries to provide free access to computers and the Internet for all ...
    (soc.culture.romanian)