RE: [AL] simple Allegro

[Derek V. T. <De...@i-...>]

Agreed, but often optimization comes at the expense
of readability. 

Huberts Law - The more you optimize a function, the
less you'll understand it a year down the line.


	A span short of a bridge.
..:: Derek Van Tonder ::.. 
..:: de...@i-... ::..


-----Original Message-----
From: Paul Fjeld [mailto:pf...@to...]
Sent: Friday, August 31, 2001 1:35 PM
To: al...@ca...
Subject: Re: [AL] simple Allegro


Hi Alleister,

You're right of course, you need to start somewhere. My
point is that what is being taught in many beginner's
tutorials for "high performance" graphics isn't the high
performance way of using the code. Why don't they start with
"cache" and "semaphore" or display list right away. You'll
>never< use OpenGL without display lists. If all you need it
for is a spinning texture mapped cube, you're wasting alot
of time. I got through OpenGL Superbible and learned how to
make a scene that >crawled<.

There must be hundreds of public 3d engines out there all
doing the same thing in nearly the same way but I bet none
of them use the "secrets of the real game gurus" (like
Unreal or Quake). There is a best practice in every field.
It evolves. It should be taught from the beginning.

BTW: my spaceflight simulator is a lunar lander - lots of
terrain CLOD ROAMing stuff. Slow unless done properly, fast
to "hint" at. Once I've abandoned the big

Take care,

Paul


----- Original Message -----
From: "Alleister" <All...@sn...>
To: "Paul Fjeld" <al...@ca...>
Sent: Thursday, August 30, 2001 7:10 PM
Subject: Re: [AL] simple Allegro


> Hello Paul,
>
> there is one good reason to still do simple 3d stuff...
> you will not one evening decide:
> "geez i think i want to be a second John Carmack right
now" and start
> writing a Q3 engine... even carmack had to start somewhere
and is
> still learning, like every other programer... if you look
at the game
> market you'll recognize that weaker 3d coders even do
comercial
> sucessfull games
>
> apart from that... a space combat simulator isn't the most
difficult
> thing to do... hardly any scenemanagement, lots of empty
space...
> nehe tuts should be enough, IMHO
>
> --
> Best regards,
>  Alleister
mailto:All...@sn...
>
>
>
>From <all...@ca...> Fri Aug 31 09:49:11 2001
Received: from hagbard.io.com [199.170.88.13] by canvaslink.com with ESMTP
  (SMTPD32-6.00) id A45FE660304; Fri, 31 Aug 2001 09:42:55 -0400
Received: from localhost (fenix@localhost)
	by hagbard.io.com (8.11.2/8.11.2) with ESMTP id f7VDd6v31659
	for <al...@ca...>; Fri, 31 Aug 2001 08:39:06 -0500
X-Authentication-Warning: hagbard.io.com: fenix owned process doing -bs
Date: Fri, 31 Aug 2001 08:39:05 -0500 (CDT)
From: Jason Wilkins <fe...@io...>
To: "'al...@ca...'" <al...@ca...>
In-Reply-To: <59E4732FAD82D511A39E0001029EA74E012084@I-WITNESS>
Message-ID: <Pin...@ha...>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Subject: [AL] Namespace Problems
Precedence: bulk
Sender: all...@ca...
Reply-To: al...@ca...
X-UIDL: 300937855
Status: O
Content-Length: 1188
Lines: 30

Just so I get my .02 cents in on the namespace collision discussion.

I think that all the functions and structs in the library should be
prepended with a (preferably short and/or configurable) id.

That this should be done for real, not as some preprocessor hackery which
renames new names to old.

That allegro.h should hack the new interface to look like the old for
legacy programs.

New programs should use a new header file, say allegro4.h

That all this should only be done after very careful consideration of the
consequences and testing.

I would also prefer to see Allegro 4.0 released before this happens.
Seriously, why propose such a huge change now?  I think that Allegro
should have been feature frozen, debugged, and released as Allegro 4.0
years ago.

I think if we do this it should be as a well thought out and tested
change, perhaps including other changes to the library which would break
the interface.  It should not be one that is rushed so that in can be a
part of 4.0 and leaves us wishing we had gone ahead and broken more of the
API while we where at it.

Oh well, I have not been following the discussion carefully enough, so I
might be completely off base.