Doom & WATCOM C
Paul Hsieh
version of the Doom alpha to an ftp site. But I must say that I am impressed
that they are using WATCOM C. I too use it, and know it to be a FAR (really
really far) superior compiler than Borland's or Microsoft's. So Id Software
has clued in. When are the rest of you? :) Seriously, why do so few
people use WATCOM C in favour of Borland C?
I know from actually using both WATCOM C 386 and Turbo C, that there is no
contest. For me, (and from what it sounds like, many of you), .EXE speed is
important. In some test that I performed Watcom C never performed at less
than twice the speed and in some cases executables went 5 times as fast.
(Yes I turned optimizing on in both). This is no joke. WATCOM C IS that
much better. Nevermind waiting for the Pentium, if you want your code to
fly 2-5x as fast right now, just get Watcom C.
If I want an integrated environment, then I use Windows or Desqview. I will
admit Turbo C's F-keys and on line help is useful occasionally, but then
again the environment gobbles up memory and I can use a reference book just
as easily, and lets face it, for experienced C programmers, all the glitzy
stuff in Turbo C is not all that useful. (I am assuming Borland C is not
that different) That "Project" idea has to be the absolute dumbest. Make
files are the only way to go.
As for compilation speed, of course Turbo is faster. Maybe many factors, but
this is irrelevant when using make. On my 486/33, WATCOMC will compile a
1000 line program in less than 30 seconds, which is good enough for me. My
modules are usually much smaller.
While Watcom's debugger may not be integrated, it is much more powerful.
Providing assembly level debugging even in protected mode. The Watcom
profiler is also quite nice, though it doesn't give me a pie chart like
Borland's apparently does (whoopie).
Basically what I am saying is that there is no substitute for object code
speed. Id knows it, I know it ... what about the rest of you? I know that
at least SOME of you are experienced enough that the trade off of speed for
mostly unneccesary gadgets is worth it. So I put the question to you. Why
not use WATCOM C?
PLUG: Watcom C 386 is a protected mode compiler for 386+ computers. It comes
with a large number of tools and libraries. With it you can create code for
Extended DOS (royalty free extender comes with it), Windows, OS/2, Penpoint,
AUTOCAD, and a few others that I cannot think of right now. It produces very
tight and fast object code, using global optimizations with a code generator
that has been fined tuned for almost a decade.
Cheers,
Paul Hsieh
chris phillips
: I was sad to hear that somebody was unethical enough to upload an unassembled
^^^^^^^^^
: version of the Doom alpha to an ftp site. But I must say that I am impressed
: that they are using WATCOM C. I too use it, and know it to be a FAR (really
A
Gee, how did you know they are using Watcom C?
Sarcasticly Yours
ChrisA
A
A
Raoul Said
I like the IDE environment. What is wrong with the project
approach anyway. I always thought make files were a pain
in the butt. Other than that it's an attractive package.
Brian Hook
> I was sad to hear that somebody was unethical enough to upload an unassembled
> version of the Doom alpha to an ftp site. But I must say that I am impressed
> that they are using WATCOM C. I too use it, and know it to be a FAR (really
> really far) superior compiler than Borland's or Microsoft's. So Id Software
> has clued in. When are the rest of you? :) Seriously, why do so few
> people use WATCOM C in favour of Borland C?
Watcom C/C++ has a lot of things going for it, most of which deal with code
generation and the 32-bit protected mode capabilities. This, however, is
not enough to make it the compiler of choice for everyone. It isn't nearly
as well supported in the marketplace as either Borland or Microsoft, and
possesses no IDE to speak of.
> I know from actually using both WATCOM C 386 and Turbo C, that there is no
> contest. For me, (and from what it sounds like, many of you), .EXE speed is
> important. In some test that I performed Watcom C never performed at less
> than twice the speed and in some cases executables went 5 times as fast.
> (Yes I turned optimizing on in both). This is no joke. WATCOM C IS that
> much better. Nevermind waiting for the Pentium, if you want your code to
> fly 2-5x as fast right now, just get Watcom C.
Have you used Borland C++ 3.1? Sure, .EXE speed is important, but so is
overall productivity. Watcom doesn't provide a Windows programming
environment which, while not that important to me, is important to a LOT of
programmers out there.
> If I want an integrated environment, then I use Windows or Desqview. I will
> admit Turbo C's F-keys and on line help is useful occasionally, but then
> again the environment gobbles up memory and I can use a reference book just
> as easily, and lets face it, for experienced C programmers, all the glitzy
> stuff in Turbo C is not all that useful. (I am assuming Borland C is not
> that different) That "Project" idea has to be the absolute dumbest. Make
> files are the only way to go.
Windows and Desqview don't provide quite the same level of functionality as
the IDE. Memory being gobbled up isn't a consideration on a machine with
8MB. And I'm a VERY experienced C/C++ programmer and I'd hate not using
BC's built in capabilities -- the instant reference manual, ease of option
selection, etc. And that "Project" idea is one of the reasons Borland's
compiler has outsold the competition consistently for many years. Make
files, while nice, are a bit on the antiquated side and are not standard
across platforms -- learn MAKE on a Sun and it won't be the same as
Borland's MAKE or NDMAKE or GNUMAKE.
> As for compilation speed, of course Turbo is faster. Maybe many factors, but
> this is irrelevant when using make. On my 486/33, WATCOMC will compile a
> 1000 line program in less than 30 seconds, which is good enough for me. My
> modules are usually much smaller.
Well, I have some 16000 lines of source code, and Borland takes about 5
minutes to compile it. If Watcom takes longer, I don't want it, at least
in the early development stages. For prototyping and debugging, Borland
wins, hands down. For final code generation I probably will use Watcom.
> While Watcom's debugger may not be integrated, it is much more powerful.
> Providing assembly level debugging even in protected mode. The Watcom
> profiler is also quite nice, though it doesn't give me a pie chart like
> Borland's apparently does (whoopie).
Borland's debugger is worlds ahead of Watcoms. And both are no where as
good as Soft-ICE.
> Basically what I am saying is that there is no substitute for object code
> speed. Id knows it, I know it ... what about the rest of you? I know that
> at least SOME of you are experienced enough that the trade off of speed for
> mostly unneccesary gadgets is worth it. So I put the question to you. Why
> not use WATCOM C?
Why not? Because although you seem to be omniscient you don't realize that
Watcom isn't all that productive in some situations. Sure, if you use
Emacs under OS/2 and want a compiler, get Watcom. But if you like the
Borland IDE (and I do) then Watcom loses for general development purposes.
I'm pretty damned experienced, and I think my judgment is pretty valid in
this area.
>PLUG: Watcom C 386 is a protected mode compiler for 386+ computers. It comes
>with a large number of tools and libraries. With it you can create code for
>Extended DOS (royalty free extender comes with it), Windows, OS/2, Penpoint,
>AUTOCAD, and a few others that I cannot think of right now. It produces very
>tight and fast object code, using global optimizations with a code generator
>that has been fined tuned for almost a decade.
As you said "whoopee!". Penpoint and ADS support -- these are niche
markets, nothing to really brag about. As I said, the Watcom compiler has
nice code generation, but lack of a decent environment rules it out for
serious development of large projects. It should be nice for final code
generation though.
Brian
PS Plus it took them some 4 years to come out with a C++ compiler after
Borland did.
--
//----------------------------------------------------------------------------
// Brian Hook "Stop! Stop in the name of all that
// ( b...@beach.cis.ufl.edu ) which does not suck!" - Butthead
//
// Mail me if you program 3d graphics and want to discuss related topics
//----------------------------------------------------------------------------
Joshua Dinerstein
> Seriously, why do so few people use WATCOM C in favour of Borland C?
Cost. All the adds that I see show Watcom C at about $900. While as a
student Borland's C products are $100 and change. Thats a big difference!
Joshua
SL...@cc.usu.edu
> Paul Hsieh
--
**************************************************************
* * *
* Joshua Dinerstein * Nothing matters and what if it did? *
* SL...@cc.usu.edu * - Who Knows!!! *
* * Anyone?? Anyone?? *
**************************************************************
If it's a raincoat why does it have to be dry cleaned?
- Some Guy on some comody show
Tim Brengle
Watcom C 9.0 and performance is outstanding.
It is the only affordable protected-mode compiler whose tools are worthwhile.
Tim Brengle
Electronic Arts
Dean Gaudet
Brian Hook <b...@beach.cis.ufl.edu> wrote:
>In article <1993Oct8.0...@math.toronto.edu> hs...@math.toronto.edu (Paul Hsieh) writes:
>> As for compilation speed, of course Turbo is faster. Maybe many factors, but
>> this is irrelevant when using make. On my 486/33, WATCOMC will compile a
>> 1000 line program in less than 30 seconds, which is good enough for me. My
>> modules are usually much smaller.
>
>Well, I have some 16000 lines of source code, and Borland takes about 5
>minutes to compile it. If Watcom takes longer, I don't want it, at least
>in the early development stages. For prototyping and debugging, Borland
>wins, hands down. For final code generation I probably will use Watcom.
Just FYI, there is an utility supplied with Watcom C that will convert
Watcom debugging information into Borland style...
>> While Watcom's debugger may not be integrated, it is much more powerful.
>> Providing assembly level debugging even in protected mode. The Watcom
>> profiler is also quite nice, though it doesn't give me a pie chart like
>> Borland's apparently does (whoopie).
>
>Borland's debugger is worlds ahead of Watcoms. And both are no where as
>good as Soft-ICE.
Depending on how you define "worlds ahead" this statement is both true and
false. WVIDEO isn't as "pretty" as Borland's debugger, but it's more
powerful if you spend a few hours learning it.
>As you said "whoopee!". Penpoint and ADS support -- these are niche
>markets, nothing to really brag about. As I said, the Watcom compiler has
>nice code generation, but lack of a decent environment rules it out for
>serious development of large projects. It should be nice for final code
>generation though.
I work almost entirely in Unix environments and I don't miss anything that
an IDE has to offer. This is a fairly religious issue however (akin to
the vi vs. emacs debate).
>PS Plus it took them some 4 years to come out with a C++ compiler after
>Borland did.
So? You can bet that Watcom's compiler will implement exactly the
proposed standard, and produce better code than Borland's. The only
argument you have remaining is the IDE vs. non-IDE argument. And for
many of us, that's a non-argument.
Dean
P.S. I feel obliged to mention that I used to be afiliated with WATCOM.
caveat emptor.
Brian Hook
> Just FYI, there is an utility supplied with Watcom C that will convert
> Watcom debugging information into Borland style...
But I doubt TD386 debugs protected mode programs.
> Depending on how you define "worlds ahead" this statement is both true and
> false. WVIDEO isn't as "pretty" as Borland's debugger, but it's more
> powerful if you spend a few hours learning it.
How is it more powerful?
> I work almost entirely in Unix environments and I don't miss anything that
> an IDE has to offer. This is a fairly religious issue however (akin to
> the vi vs. emacs debate).
Well, under Unix you can have multiple shells open with Emacs in one,
compiler in another, test in another, etc. or use M-x compile or something.
This is as good as the IDE, but it takes a bit longer to set up (getting
context sensitive help, for example).
>PS Plus it took them some 4 years to come out with a C++ compiler after
>Borland did.
> So? You can bet that Watcom's compiler will implement exactly the
> proposed standard, and produce better code than Borland's. The only
> argument you have remaining is the IDE vs. non-IDE argument. And for
> many of us, that's a non-argument.
If it takes Watcom that long to come out with things, I'd rather stick with
someone on the cutting edge. The IDE is not for everyone, but for the
majority of users I would say they REALLY like it. Specifically, it's a
good editor (not quite Emacs, but good enough) with nice features, edits
fairly large files, and, most importantly, does things like allow for
project management, takes you directly to errors, allows adjust options via
menus, allows integrated debugging, and has online help with sample code.
These are HUGE boons to productivity, at least for me. Then again, some
people prefer using Emacs w/ GCC and GDB, and I can't begrudge them that.
But on a PC or DOS environment, the IDE makes life a LOT simpler. MAKE
just doesn't cut it when prototyping code. Or changing options.
Brian
Philip Brown
In <BWH.93Oc...@beach.cis.ufl.edu> b...@beach.cis.ufl.edu (Brian Hook) writes:
>If it takes Watcom that long to come out with things, I'd rather stick with
>someone on the cutting edge.
Really? I'd rather stick with a product with no bugs. Borland C++ doesn't
handle class initializers properly under Windoze.
(They don't get called at all.. Damn silly)
--
--------------------------------------------------------------------
"I want World Peace"
"We're the US Government. We don't DO that sort of thing"
ph...@cats.ucsc.edu ph...@soda.berkeley.edu
sl...@cc.usu.edu
> ^^^^^^^^^
> : version of the Doom alpha to an ftp site. But I must say that I am impressed
> : that they are using WATCOM C. I too use it, and know it to be a FAR (really
> ^^^^^^^^
> A
> Gee, how did you know they are using Watcom C?
>
> Sarcasticly Yours
Well, I knew for three reasons: Jay told me months ago, I have the Doom Alpha,
and I recognized the source generation techniques used in the disassembly!
Joshua Jensen
sl...@cc.usu.edu
>
> Cost. All the adds that I see show Watcom C at about $900. While as a
> student Borland's C products are $100 and change. Thats a big difference!
>
> Joshua
> SL...@cc.usu.edu
Josh, you know as well as I do that mail order companies sell it for $345 > or
< by a few dollars.
Joshua Jensen
sl...@cc.usu.edu
>> While Watcom's debugger may not be integrated, it is much more powerful.
>> Providing assembly level debugging even in protected mode. The Watcom
>> profiler is also quite nice, though it doesn't give me a pie chart like
>> Borland's apparently does (whoopie).
>
> Borland's debugger is worlds ahead of Watcoms. And both are no where as
> good as Soft-ICE.
Have you l9ooked at Watcom's debugger? It is WORLDS ahead of Turbo Debugger
any day. It may not be on the surface, but take a good look at the help and
the manuals.
> As you said "whoopee!". Penpoint and ADS support -- these are niche
> markets, nothing to really brag about. As I said, the Watcom compiler has
> nice code generation, but lack of a decent environment rules it out for
> serious development of large projects. It should be nice for final code
> generation though.
And Pentium support and others...
Joshua Jensen
Tim Triemstra
that it is now about $395 from Programmer's Paradise (the C++ v9.5 too!)
I personally will stick with Symantec C++ but that's me. I like the fact
that it will use Borland and MSC style object code and it creates Win32s
and it has a better Windows hosted IDE that any of the afore-mentioned
products. However, to each his own (I've used at least 5 different
compilers and never stuck with any of the provided platform systems.)
Tim Triemstra
University of Michigan
emp...@ais.org
Jay Stelly
>In article <1993Oct8.0...@math.toronto.edu>, hs...@math.toronto.edu (Paul Hsieh) writes:
>> Seriously, why do so few people use WATCOM C in favour of Borland C?
>
> Cost. All the adds that I see show Watcom C at about $900. While as a
>student Borland's C products are $100 and change. Thats a big difference!
>
> Joshua
> SL...@cc.usu.edu
>
>> Paul Hsieh
I'll second that. I'd like to have Watcom at home, but I'm not about to pay
that kind of money for it. The lowest end Borland C++ (maybe Turbo C++) can
be bought for $69.00 US. Borland is sold through many more dealers and is
thus more heavily discounted.
Jay Stelly
j...@cs.com
Brian Hook
> As you said "whoopee!". Penpoint and ADS support -- these are niche
> markets, nothing to really brag about. As I said, the Watcom compiler has
> nice code generation, but lack of a decent environment rules it out for
> serious development of large projects. It should be nice for final code
> generation though.
> And Pentium support and others...
Pentium support is nice, but is it CRUCIAL? There are what, 10 times as
many (or more) 486 systems being sold every month as Pentiums. Hell, I've
only seen ONE Pentium system, and I'm around hundreds of computers every
day. Not quite a consideration yet, and Borland'll have a new compiler out
long before the Pentium gets up steam.
Brian
Brian Hook
> Josh, you know as well as I do that mail order companies sell it for $345.
Clue: students usually don't have that extra 245 bucks available to blow
on a compiler.
Dr. Cat
3.0 installed on my hard drive to have Turbo Debugger around, which wasn't
included with the 2.0 I got. I have a complete set of disks and manuals for
3.1 lying around, but I'm too lazy to install them.
What do I think of Watcom? I'm sure it's very nice for a lot of people, but
I don't need it. The few routines in my code that burn up the majority of
the execution time generally get recoded into assembly language. The
remaining code is responsible for a small enough percentage of overhead that
the speed increases gained from switching to Watcom would be pretty
negligible. I like project files and online help, too. Project files are
better than makefiles for me because I was able to just grab my mouse in the
Borland IDE and >>GUESS<< how to set up, compile, link and run something from
multiple source files, and it worked right the first time. I tried to make a
makefile once by taking one of someone elses and modifying it, and it didn't
work. So I'd probably need to read a manual to learn how, spending time I
could be using to make up some clever puzzle or something.
I'm sure Watcom is a more powerful tool than what I'm using. But I don't
NEED a more powerful tool. Any C compiler on the market is a far more
powerful tool than the 6502 assemblers I cut my teeth on, and I'm perfectly
happy with all the ones I've used. I stick with Borland for the IDE. 'cause
I'm not really concerned with getting a bigger hammer, when they're all big
enough to crack open walnuts, and I'd rather just get the one with the nice,
soft, furry stuff on the handle to keep my hand comfortable.
Dr. Cat / Dragon's Eye Productions
sl...@cc.usu.edu
> > As you said "whoopee!". Penpoint and ADS support -- these are niche
> > markets, nothing to really brag about. As I said, the Watcom compiler has
> > nice code generation, but lack of a decent environment rules it out for
> > serious development of large projects. It should be nice for final code
> > generation though.
>
>> And Pentium support and others...
>
> Pentium support is nice, but is it CRUCIAL? There are what, 10 times as
> many (or more) 486 systems being sold every month as Pentiums. Hell, I've
> only seen ONE Pentium system, and I'm around hundreds of computers every
> day. Not quite a consideration yet, and Borland'll have a new compiler out
> long before the Pentium gets up steam.
This thread, as does every other one on rec.games.programmer has become
completely pointless. Get DOOM out of the subject line and move it to
comp.lang.c++.
Joshua Jensen
sl...@cc.usu.edu
>
> Clue: students usually don't have that extra 245 bucks available to blow
> on a compiler.
Are you one of those GUS fanatics either for or against the GUS? That's
exactly what you're sounding like.
Joshua Jensen
Geoff Thomas
Well, I'm going to get Watcom, just as soon as I can afford, it (which is pretty
soon), because I'm fed up to there-and-a-bit-further with 64k arrays, and progs.
less than 640k, and I can't be bothered hard cording in garbage to use E/XMS and
rubbish like that.
I'd stick with Borland (currently using Turbo C) if they brought out a 32-bit
comp. 'cos I DO like the IDE, but this memory stuff is too much of a nuisance.
Someone mentioned about using different sized hammers to crack walnuts; I want to
crack large bloacks of concrete! And Borland just isn't there.
Geoff Thomas
Alan Young
I picked it up through mail-order for $129+shipping. Now, if memory
serves, Borland started out with Turbo C somewhere around this price
several years ago. We might be seeing the start of a 386 compiler
price war. Now if they can just shrink the extender .EXE size... :)
Alan
-----------------------------------------------------------
Alan Young | Are you able to follow the Red Team ?
ayo...@holonet.net | -Front242
-----------------------------------------------------------
Leo Lipelis
>majority of users I would say they REALLY like it. Specifically, it's a
>good editor (not quite Emacs, but good enough) with nice features, edits
>fairly large files, and, most importantly, does things like allow for
>project management, takes you directly to errors, allows adjust options via
Borland's editor the one that is built into IDE makes me PUKE!!
I mean it's pathetic, wimpy, featureless editor. Heck, vi has
more features. Brief is kind of nice, but Emacs 19 makes it blush.
I wish I had Emacs 19 port for 386 that would run in 2-4 megs. ;)
But I spend most of my time in Unix, so I don't really care all that
much.
--Leo.
Dr. Cat
done, and there's >21< new posts on this topic. The most any other subject
has is 3! Do people find talking about the process of making games to be
less interesting than talking about which compiler to do it with? I hope
not.
Someone brought up the price issue, and I thought I'd mention another price
nobody has said anything about. $0. Yup, absolutely free. It's the Ihaveit
1.0 C compiler. 'Cause I have to believe the vast majority of people here
already have SOME kind of C compiler (or maybe Pascal, if they prefer it) on
their machine that they've been coding with already. Maybe you paid for it
some time back, but the point is, you have it now, and can already write C
programs without paying additional money to anybody for a new one.
It costs something else to get a new C compiler too, aside from $100 or $345,
or $666, or whatever you might spend. It costs time. If somebody tells me
that Super-Ultra-Splendid C version 14.0 has a magnificent artifical
intelligence based install program, that simultaneously sets it up on your
hard drive AND tutors you in all the commands and parameters necessary to have
you as fluent in it as you were in your old compiler in just FIFTEEN MINUTES,
I still have to ask - have you finished your first game yet? Because all I
can say is MAYBE Super-Ultra-Splendid C version 14.0 will speed up your
development cycle enough to more than make up for those fifteen minutes and
save you time, or maybe it won't. But I GUARANTEE you, if you spend those
same fifteen minutes working on your first game, you'll get a little bit
closer to being done than you were before. As for whether the new compiler
could make your game better, whether it makes you finish sooner or later,
trust me on this one. The first game you ever make will NOT be the best game
that the hardware you work on is capable of supporting. It will NOT be the
best game that you yourself are capable of making. It probably won't be
anywhere CLOSE to either of those things. But your second game will be
better for what you learned from it. And yet, how can you ever make your
second game without having made your first game? Short of inventing a new
way of counting to two, it's pretty tough.
So here's my personal advice on which compiler (or interpreter, or assembler)
to use to make your first game. Anybody who's working on their second, or
fifth, or ninety-seventh, feel free to ignore this and spend as much time
picking out and/or discussing better tools. But for your first game, use the
compiler you have on your machine already. I promise you, it is possible to
make functional games with just about any C compiler out there. It's one
less thing to worry about. If you feel the need for a better compiler for
your second game, switch then. Not now. If you keep looking for the best
tools, I'll know you'll be out there upgrading to Dos 8.0 a few days after it
ships, instead of spending another valuable fifteen minutes working on your
game, and you'll never finish it. Not until after you get that new &
improved shareware caching program installed, and, and... I can't
overemphasize the value of finishing that first game, either. The warm
feeling of looking at it on your screen and knowing it's done, of watching
over someone's shoulder as they try it out, and knowing you made it... The
satisfaction of finishing things is what makes it possible to continue to do
something this difficult as a career, over and over again. And it doesn't
come in any box from Borland, Microsoft, Watcom or anyone else, this whole
"finishing things" business. It has to come from inside of you. If you've
got it, you'll be able to use it no matter which compiler you have. And if
not - well then it hardly matters which compiler you're using, does it?
Brendan Jones
>b...@beach.cis.ufl.edu (Brian Hook) writes:
>
>>majority of users I would say they REALLY like it. Specifically, it's a
>>good editor (not quite Emacs, but good enough) with nice features, edits
>
>I mean it's pathetic, wimpy, featureless editor. Heck, vi has
>more features. Brief is kind of nice, but Emacs 19 makes it blush.
>I wish I had Emacs 19 port for 386 that would run in 2-4 megs. ;)
Featureless? Nah! It has most (all?) of the Wordstar compatible
commands. It has global search and replace, regular expression
support, autoindent, you can have multiple files open with multiple
windows for the same file, keyword highlighting, inbuilt error
message tracking, markers, clipboard support. Featureless?
>But I spend most of my time in Unix, so I don't really care all that
>much.
Yes, I was once a Unix biggot too ;-)
>
>--Leo.
cheers
bj
Howard Louie
it was some well put, practical advice. I always remember that when
I start programming in the "real" world. thanx.
William Hoffman
>to use to make your first game. Anybody who's working on their second, or
>fifth, or ninety-seventh, feel free to ignore this and spend as much time
>picking out and/or discussing better tools. But for your first game, use the
>compiler you have on your machine already. I promise you, it is possible to
>make functional games with just about any C compiler out there.
Or Pascal. :)
(Ducking to avoid buckets of flaming pitch)
George