Follow Slashdot stories on Twitter

 


Forgot your password?
Close
typodupeerror
×
News

Dave Taylor Interview 47

Posted by CmdrTaco from the stuff-to-read dept.
Mr.E writes "Dave Taylor, former id code monkey and the man responsible for the linux port of Doom, is interviewed by Ga-source about the past and future of Linux gaming." He talks about tons of stuff from Id to Crack.com to Transmeta (no he doesn't spill the beans). Good bit.
This discussion has been archived. No new comments can be posted.

Dave Taylor Interview More

Dave Taylor Interview

Comments Filter:
  • According to the change log [cygnus.com], a question [cygnus.com] and another question [cygnus.com] on the egcs mailing list, exceptions are thread-safe.

    Oh well, I'll soon find out for myself.

  • Well, er... Seems fairly straightforward.

    I've been thinking for quite a while now that sooner or later, we're going to see a lot of GPL'd (or similarly licensed) game development. There's an increasing trend lately towards licensing game engines developed by other companies, simply because it seems to be a lot less work than developing one from scratch.

    If a really high quality FPS engine, for example, gets GPL'd, I can see people using it for projects and contributing the code they write back to the engine, and selling the finished game as a product by keeping the art/level design/etc. as their own property. Or am I mistaken in thinking this could be done within the terms of the GPL?

    This would especially make sense given how much it must cost to produce a commercial quality game these days...

  • I seem to recall when crack.com folded Dave updated his .plan file with a statement to the fact that he was about announce his next project. It was going to revolutionize the world/internet and be prepared to be wowed.

    The world/internet still seems the same to me. Was he talking out of his ass or did he mean he was going to work for Transmeta? If it was, just what the heck are they up to?

    ps. I could have this statement mixed up with other .plan monkey but I think it was D. Taylor's.

  • Do what Torvalds does: look at the assembly output.

    I think you'll find that GCC does NOT generate better code. I'm not saying it's a bad compiler, but you got to remember that this is a portable, cross-platform, cross-target compiler. MSVC is optimized for Wintel.

    I've been using MSVC 4.2 to compile an in-house OS for a long time now just because the optimizer kicks ass. Since 5.0 had this bug in the optimizer (that I've never seen confirmed to be fixed) I've never trusted the later versions though.

    As with any compiler, to have it turn out nicely optimized code you _have_ to look at the assembly output from time to time. Some functionally identical (C) code just generates better assembly then other (and it could be opposite between compilers - hmm, I wonder if that could be used for an other Mindcraft debacle)

    Breace.
  • I buy 10,000 shares of Redhat, or 10,000 shares of 'Linux.Business.whatever.inc' and now I have the incentive to help make laws that protect those companies. If the average 'suit' doesn't have a stake in GPL_using companies, he doesn't have a stake in insuring that the GPL is soundly legal and well supported.

    For instance: Someone sues Redhat over patent infringement on a GPL'ed program they distribute.
    They sue Redhat because Redhat has money, as opposed to FSF or Debian who have nothing. As a Redhat Stockholder, it is in my selfinterest to get laws passed that protect Redhat's position.
    Thus I will help get GPL protecting laws passed.
    And fight software patent laws at the same time.

    Dave's point on why they don't use GPL'd code right now: Because the legal shakiness means people don't want to put all their eggs in one basket. Try convincing your boss to let you use GPLed code inside your product, and watch how the legal aspects tied everything up in knots. It's not that the GPL is so bad, it's that people don't trust it because it's legally untested and unprotected still.

    P.S. Substitute BSD, Artistic, other open_license for GPL as you wish. Do not substitute APSL, NPL, other_barely_Open, as you wish. :)

    p.p.s. I am a big fan of GPL. I personally think it'll stand up court, should it ever come to it.
    Your lawyer_droids might disagree. If so, please send them back the factory for reprogramming.
  • And device enumeration... is it _really_ needed? How many people actually have more than one display device? (Add ons like Voodoo not included...) Why wouldn't I want to use the primary device?

    Did you READ the article?

    Dave talks about how the future is in multiple monitors. 2 17 inchers beats 1 21 incher any day.

    Place 3 monitors in front of you, and bingo: widescreen MEGA Game. Turn the outer 2 slightly toward you, and you have amazing feel of realness due to the extra degrees of field of vision. Add more monitors as your CPU/3D/budget/wife allows.

  • Are multiple monitors usable in Linux?

    --

  • And gee, who worked on Doom? Why, it was Dave Taylor. :)

    I've often wondered why more games didn't support multiple screens like this. I can think of few games that were too crowded on one monitor to play.

    I expect that the cheaper computers get, and the more Linux and other 'networked from the ground up' OSes spread, the more we will see games that be run off of multiple screens/cpus/videocards...
  • Is there any reason why I get nothing but a
    blank window when I clink on http://www.ga-source.com/interviews/ddt.shtml

    ?
  • What Linux needs is more developers. It has a never-ending thirst for them. The more you get, the more gems you get, the more likely the tough, killer apps, drivers, and infrastructure are developed.

    well i think the third-party tools developers are also hoping for this as well. not long after the Linux is Not Red Hat (red-hat/codewarrior) [slashdot.org] article, i got an email from mark [mailto] of metroworks customer service asking a few questions and agreed to a comment i made about the 'new breed' of application developers that will eventually flood linux.

    ... linux distributions such as 'rh', suse and caldera (and i single these distro's out because they are targeting the business market..) gain more market share the developers will more likely be 'application-developers' rather than the traditional hard core system hackers typified by the comment, 'true coders used Debian and Slackware'....

    it was good to read another ddt rant (informative and easy to read with a bit of a chuckle)..things got pretty boring on finger.planetquake after he left. [planetquake.com]
  • I think it was Jonathan Clark who said that.
  • Getting Cavedog to port Total Annihilation!

    Imagine playing that game with SMP!
  • The license of the engine isn't the point. I remember reading an article (I can't remember the source) comparing licensing the Unreal and Quake ][ engines. The editors in question recommended the Unreal engine, if only because it was better supported. The attitude from iD Software seemed to be "Here's the code, have fun", where Unreal came with copious hand-holding.

    It's almost a pity that the games on the Unreal engine ended up sucking.

  • Try turning off your Java und JavaScript. Worked for me.
  • OK, I admit making assumptions about how something is going to be used is stupid.

    As for supporting devices that don't exist, I was thinking more of testing. How do we test how our software works on hardware that is still on the drawing board or a gleam in the designer's eye.
  • It's the guy from Vault 13... damn small monitors.. ;)

  • Great Quote. (Score:3)

    by BigEd ( 6405 ) <togdon@thanatoid.net> on Saturday June 19, 1999 @08:41AM (#1842473) Homepage
    "I think trying to direct Linux is like herding cats. Cats are not motivated by anything resembling what motivates you, but if you get enough cats, some portion of them will go the right direction just by chance."
  • I just got home from 16 hours networked Warcraft with 2 other guys ... I don't like StarCraft ( or any space based games ) but just can't get enough of Warcraft. It is the perfect game for me ...
  • Oh wait wrong story, I meant to say I can't wait to find up what the hell transmeta is up to. Espically with all that shit he just said about it. I don't know if their product will be able to live up to all the hype.
  • GCC is _not_ superior to MSVC ! Well, it is when you need to write code for 10 platforms but from technical point of view gcc-generated code is much worse ... Wake up dude. Eveybody uses GCC cause it is evailable for just about every platform out there and , well, there is nothing else available for Linux ( maybe KAI ... is it better then GCC ? Anyone ? )
  • The most shocking thing to me is the how Dave talks about how wonderful open source is, and how great Linux is because of that; but on the other hand completely accepts the insane secrecy at Transmeta. Now, I suppose I'd give my eyeteeth to work there too, but it's an odd juxtaposition that is made odder in that he doesn't seem to notice it himself.

    The description of programming for windows is great. I'd had exactly that experience; it's upsetting that OpenGL programs run so much faster under windows than Linux. Still, for the FX work that we do, it's much more important to have real operating system support than fast graphics.

    And finally, I agree that the integration of X window system support for games is extremely important; not for the success of the games as for the success of the X window system. If we make X work for games, it will work for anything -- and there will be cards and other hardware that will almost magically appear due to the awesome economic power of the gamers.

    thad
  • If you happen to have a matrox card, you can use matroxfb to wake up a secondary matrox card. I'm doing so right now.. You should use Gerd's fbdev patches as well since the normal mga driver assumes that the matrox card is the primary card.
    I needed to remove the vt switching code from the other server as well, but that was pretty straight forward.. and I'm using x2x to move the pointer between the screens.

    Hope that made any sense... anyway you could just wait for XFree86 4.0 which is going to have multihead support. Or you could buy a commercial X server with multihead support.

    /Andreas Ehliar
  • I'm pretty sure that MSVC compiles all of the requisite header files into an intermediate file, (if you use MSVC, you see *.pch files in with the rest of your object files - that be them), and so instead of recompiling and repreprocessing every one of the massive Windows header files every time you hit the compile button... well, you get the point.

    There are probably other techniques involved as well, but I know that one is there.
  • I have a nice, medium-sized C++ project with around 15 KLOC. This project does not use any kind of GUI, but it does have to include all the windows.h bloat on NT. Using the WIN32_LEAN_AND_MEAN switch helps a little, but not much.

    I don't know how representative this is, but egcs 1.1.2 on Linux compiles and links this project almost twice as fast as MSVC++ 5.0 on NT 4.0. In general, I was surprised how fast egcs can be, because I used to think of gcc as a nice compiler, but slow as molasses.

    Also, precompiled headers and incremental linking don't always seem to buy a speedup. There's one particular project on Borland C++ Builder 3 where incremental linking slowed down build times by a factor of 2.
  • After what he said my curiosity is even greater than it was already.

    Am I reading too much into it when I note that the website we all love to hate [transmeta.com] is down. Will it be back up on Monday with some content?

    Unless they are contracting for the NSA they will have to tell the public what they are up to at some point, otherwise how can anyone buy the product.

    At the moment the best bet seems to be that it is a project rather like the Elbrus E2K [elbrus.ru], but who knows?

    When it comes online they are going to need the 4-way SMP and Apache tweaks to stop being slashdotted...

  • Somewhere in "Inside DirectX" (I can't find it at the moment), it talks about how DInput was designed for joysticks and controlors that have yet to be made. WTF? Why would I support a device that doesn't exist?

    I assume you're joking..?

    If not, imagine, say, a Flight Simulator designed for joysticks with 4 buttons via DirectInput, then you plugged a joystick in with e.g. 20 buttons (as some of them do), and your FS doesn't let you assign functions above the first 4 buttons, because the developers though "why support a device that doesn't exist?"

    And device enumeration... is it _really_ needed? How many people actually have more than one display device?

    Me. I have two at work, and am planning to get a second monitor at home.

    (Add ons like Voodoo not included...) Why wouldn't I want to use the primary device?

    Well, because of the example you just gave? I've run a couple of DirectX games that didn't use my Voodoo card because the Matrox Mystique was first in the DX list of 3D drivers. So I had to put up with sucky performance and stippling instead of alpha blending. And I didn't even have two monitors.

    Or how about that my first monitor is an LCD panel, and the persistence makes it not great for playing games, so I'd rather use my CRT, which is the second monitor - except I can't, because you don't see the point as hardly anyone has a second monitor, and 640k should be enough for anyone...(sarcasm off)

    Games ought to support display (and input device/sound device) enumeration if possible. Sometimes time pressures will force such support out, but if you have a choice, do support it.

    Tim

  • A rather good interview...but one thing didn't ring true in my head:

    "OK, here's a fairly far-sighted prediction. I think PC games in general are going to move away from being full-screen, and they're going to start being windowed."

    Gamers tend to like to immerse themselves in the gaming experience. I somewhat understand his explanation about larger screens becoming the norm, but (for me, at least) one of the most frustrating things is having to play a game in windowed mode. It tends to take away from the atmosphere, making it harder to lose track of yourself and just PTFG (Play The Fucking Game). This is true especially in adventure and role-playing games, where a good game of the genre can make you think you are Tex Murphy, or Sonny Bonds, or the dude from Vault 31...if only for a little while. With your desktop staring back at you around the borders of the game window, this is much harder to do.

    You dim the lights in a movie theater when the film starts for a reason.
  • Dave Taylor:
    "Then you read about some of the functions and you smirk because you see the half-hearted attempts to emulate Unix or X Windows, you can see the
    entomology and everything."

  • but MSVC has bigger problems... (Score:1)

    by Anonymous Coward

    Problem is, MSVC doesn't come nearly as close to compiling real C++ code and having it work. MSVC still thinks "new" should return NULL for failed allocations and that variables declared at for (...) scope should remain in scope for the rest of the function (or other enclosing scope). On the other hand, MSVC does support thread-safe exception handling, unlike the current releases of FSF and egcs g++...
  • *sigh*

    I'll go reboot the web server.

    Damn...and only 500 days uptime....

Slashdot Top Deals

Nothing ever becomes real until it is experienced. - John Keats

Close