Joel Koltner wrote:
> Good post, David -- thanks.  I don't agree completely with everything, but I 
> think your points have plenty of merit.  I'd only add that I just haven't had 
> the amount of problems with registry corruption that it appears that you 
> have -- if I had, I could very easily find myself thinking much more poorly of 
> the registry.
> 

I've only seen one windows installation with the registry as completely 
trashed as the one I mentioned (as another poster said, there are a few 
more recovery procedures I could have tried).  But partly corrupted 
registries are not uncommon - I've had my user registry scrambled a 
couple of times.

> "David Brown" <d...@hesbynett.removethisbit.no> wrote in message 
> news:M...@lyse.net...
>> There *are* standard tools for viewing and editing *nix configuration 
>> files - any text editor will do the job.
> 
> Yes, my criticisms were that the format isn't standardized (as you point out), 
> often *isn't* hierarchical, and -- for the *average* user -- isn't always in 
> an "obvious" location (copying /etc is easy, but then you need, umm... 
> .mozilla?  .gnome-desktop?  etc.? --> the average user shouldn't have to dig 
> through their home directory figuring out which "dot" directories contain 
> configuration information).
> 

The average user doesn't need to directly access configuration files (or 
the windows registry) for common programs.  To most users, it does not 
matter whether Firefox settings are in ~/.mozilla or ~/.firefox, as they 
access them with "Tools | Options".

Having said that, the location for almost all configuration files *is* 
obvious.  If it is a system level configuration file, it will be in 
/etc.  If it is a user level configuration file, it will be in a hidden 
file or directory in your home directory (it's hidden simply because you 
are not normally interested in seeing these files when you are looking 
for other files in your home directory).

>> Oh, and one more thing - *nix configuration files are almost always well 
>> documented.  How much windows software comes with documentation for the 
>> registry settings?
> 
> For publicly-developed programs, almost as many?  ...it's just that so many 
> more "typical" programs on a Windows PC are commercial, and these of course 
> often don't document their settings.
> 

If by "publicly-developed" you mean "open source", then I think you are 
probably wrong.  Windows registry settings are often considered 
"developer information", not "user information".  So they are documented 
in the same way as most programmers (open or closed source) document 
most of their code - there are brief comments, but otherwise the 
programmer assumes the information is obvious.

For windows "commercial" software (or more generally, closed source 
software - including freeware), you are right that the registry settings 
are seldom documented.  However, for commercial and/or closed source 
software on *nix, configuration files *are* generally documented.


>> This was NT 3.51.  *All* these solid design decisions were thrown out step 
>> by step through NT 4.0, W2K, XP, and Vista (against Dave Cutler's 
>> recommendations, I believe).
> 
> Yeah, I know -- ostensibly it was for performance, right?  That video back in 
> the NT 3.51 days going through the Kernel each time was never going to be fast 
> enough for games/multimedia applications?
> 

Many of the changes were for performance reasons, others were for 
backward compatibility, and others were for economic reasons - sales on 
non-x86 platforms were too low for MS to spend the time and money on the 
Alpha, MIPS and PPC ports of Windows.  Of course, if they'd just written 
the system properly, the ports would have involved only a tiny 
proportion of the source code at negligible cost.  The problem was that 
in large sections of the windows code that could easily have been 
cross-platform, programmers were assuming an x86 architecture.

AZ Nomad schreef:
> On Tue, 26 Aug 2008 14:31:30 +0200 (CEST), Jack <p...@pluto.com> wrote:
>> AZ Nomad <a...@PremoveOBthisOX.COM> wrote in
>> news:s...@ip70-176-155-130.ph.ph.cox.net: 
> 
>>> The joy of microsoft operating systems that have consistently stayed
>>> two steps ahead of available hardware.  Requiring 3 BILLION bytes of
>>> memory just to draw a desktop and run a few relatively simply low
>>> performance applications like a word processor and web browser is an
>>> incredible feat.  
>>>
>>> Just think.  In ten years will have 200ghz machines and microsoft will
>>> still make them dog slow.
> 
>> That's not (only) Microsoft. That's Software Engineer and OO programming.
> 
> It is only microsoft.  Other companies have no problem using OO
> programming and actually getting *more* efficient over time.

Only Microsoft?  You are showing you bias. Adobe is another fine 
example, ever looked at recent versions of Acrobat Reader? It gets 
buggier, slower, bigger and behaves more and more like a pig with every 
new release. And frankly even Linux distributions have become more a lot 
more bloated over the years.

I do remember the time when you could still comfortably surf the 
internet and run a wordprocessor on a machine with 64MB and a processor 
than ran at less than 300MHz, and I can't help wondering why nowadays 
you need at least 0x10 times as much to do basically the same job.

Dombo <d...@disposable.invalid> writes:
> I do remember the time when you could still comfortably surf the
> internet and run a wordprocessor on a machine with 64MB and a
> processor than ran at less than 300MHz,

I remember doing all that on a 5 MHz S-100 bus machine with 768k RAM
and a 360k floppy drive.  Of course, it wasn't called "the internet"
back then.

Let the one-upping begin :-)

On 26 Aug 2008 16:26:24 -0400, DJ Delorie <d...@delorie.com> wrote:

>Dombo <d...@disposable.invalid> writes:
>> I do remember the time when you could still comfortably surf the
>> internet and run a wordprocessor on a machine with 64MB and a
>> processor than ran at less than 300MHz,

>I remember doing all that on a 5 MHz S-100 bus machine with 768k RAM
>and a 360k floppy drive.  Of course, it wasn't called "the internet"
>back then.

And there ws no such thing as a web browser, or even a GUI.  All
applications were character based.  

AZ Nomad <a...@PremoveOBthisOX.COM> writes:
> And there ws no such thing as a web browser, or even a GUI.  All
> applications were character based.  

Well, no web browsers at least.  We had GUIs but they weren't as
sophisticated as now, and many of my applications were graphics based
- the machine came with a 640x480 graphics card by default.  It wasn't
until the IBM PC was released two years later that crappy graphics
became the norm.

Dombo <d...@disposable.invalid> writes:

> The problem with Windows is that out of the box it is pretty easy to
> screw it up because for user convenience everyone gets
> administrative privileges. You could solve that problem by removing
> administrative privileges for those users that are too smart for
> their own good. With proper user rights it can be pretty hard to
> screw up a Windows installation.

Unfortunately, with proper user rights it can also be pretty hard to
_use_ a Windows isntallation - so many apps need to be installed by an
Administrator (not just a Power User), and some even need to be _run_
thus.  So [almost] everybody does.

Back to the Subject - has anybody here used BRL-CAD[0] in anger?
(It's "just" a CSG modelling tool with Tcl and a raytracing engine, so
its best application is in the physical-layout arena, but I can see
how it could be bent to PCB layout with some little effort.)

mlp
[0] it's free, and runs on multiple platforms

Michael N. Moran wrote:
> Joel Koltner wrote:
> 
>> Unix has a similar problem with configuration files...
>> everyone and their brother invented their own format for
>> storing settings (since there's still no standard central
>> repository for settings as the registry in Windows
>> provides),
> 
> 
> Like the registry, which is *so* much better ;)
> 
>> and while most are simple enough to figure out via
>> examination of what's already there,
> 
> 
> [Using a simple text editor]
> 
>> many are somewhat "brittle" as well (the common example
>> being how easy it is to break X windows by, e.g., leaving
>> out a semi-colon in Xorg.conf ... sheesh...)
> 
> 
> Fortunately, the Windoze registry is intuitive *and* robust.
> 
>> It took years for those in the Unix world to even get
>> together on something as simple as how new programs
>> should be programmatically added to a "start" menu; happily the Gnome 
>> and KDE guys both seem to play nice on
>> this issue today.
> 
> 
> As opposed to the years that it took Mirco$oft to figure out
> how to do preemptive multi-tasking?
> 
> But I digress :-P
> 
   The *idea* of a registry seems to be OK, but in WinDoze, it is a 
veritable garbage can.
   I think that each program should have its own *findable* (and 
readable) registry in the same directory that the program is stored.
   That might make it easier to !totally! remove a program...

On Tue, 26 Aug 2008 06:51:37 +0000, Guy Macon wrote:

> Perhaps it's just me, but i expect a quad-core 3GHz machine to be able
> to keep up with my typing as well as my Commodore 128 does.

Software gets slower faster than hardware gets faster. 

c.f. Vista.

-- 
		Przemek Klosowski, Ph.D. <przemek.klosowski at gmail>

On Tue, 26 Aug 2008 22:05:06 +0200, Dombo <d...@disposable.invalid> wrote:
>AZ Nomad schreef:
>> On Tue, 26 Aug 2008 14:31:30 +0200 (CEST), Jack <p...@pluto.com> wrote:
>>> AZ Nomad <a...@PremoveOBthisOX.COM> wrote in
>>> news:s...@ip70-176-155-130.ph.ph.cox.net: 
>> 
>>>> The joy of microsoft operating systems that have consistently stayed
>>>> two steps ahead of available hardware.  Requiring 3 BILLION bytes of
>>>> memory just to draw a desktop and run a few relatively simply low
>>>> performance applications like a word processor and web browser is an
>>>> incredible feat.  
>>>>
>>>> Just think.  In ten years will have 200ghz machines and microsoft will
>>>> still make them dog slow.
>> 
>>> That's not (only) Microsoft. That's Software Engineer and OO programming.
>> 
>> It is only microsoft.  Other companies have no problem using OO
>> programming and actually getting *more* efficient over time.

>Only Microsoft?  You are showing you bias. Adobe is another fine 
>example, ever looked at recent versions of Acrobat Reader? It gets 
>buggier, slower, bigger and behaves more and more like a pig with every 
>new release. And frankly even Linux distributions have become more a lot 
>more bloated over the years.

Perhaps, linux doesn't make the fastest desktop computers run like shit.

Maybe with 10ghz of cpu and 20GB of ram vista might run OK, but otherwise
it's a steaming mountain of pig shit.

And adobe's latest reader runs just fine under linux on my 1.4 ghz laptop
with a typical 5200rpm laptop drive.  If MS wrote it, you could count on it
being unable to run on such hardware.


A $250 computer from 5 years ago will run extremely well w/ linux where
vista won't run at all.  

On Wed, 27 Aug 2008 06:05:09 GMT, przemek klosowski <p...@gmail.nospam> wrote:
>On Tue, 26 Aug 2008 06:51:37 +0000, Guy Macon wrote:

>> Perhaps it's just me, but i expect a quad-core 3GHz machine to be able
>> to keep up with my typing as well as my Commodore 128 does.

>Software gets slower faster than hardware gets faster. 

And microsoft does it faster than the chip makers can.