Digital Mathom

Epic fail by the Khronos ARB WG

2008-08-13T04:41:00.008+02:00

So the Khronos OpenGL ARB WG, after a year of almost complete, unexpected silence instead of the promised OpenGL 3.0 specification, released something that they for some reason call 3.0 but which has almost nothing of what was promised.

Instead of the promised new object model, the one that was supposedly almost done, it contains among other things a slightly modified EXT_framebuffer_object promoted to core. This extension is also known as that horrid piece of amazingly poor design that (according to at least one former ARB member and GL3 contributor) prompted the creation of the new object model in the first place.

Yay.

What bugs me the most, however, isn't even the way the 3.0 spec. turned out, but rather that their year of silence came after their move into Khronos for efficiency reasons and after heralding a new era for OpenGL with these words:

Regardless of the poll, you may find yourself asking, “Why, after all these years, is the OpenGL standards body finally opening up and sharing with its audience, its devoted developers, its enthusiastic end-users, its people?” It must be a maturity thing. It took us a solid 14+ years to shed our youthful shyness and find a voice. The last decade was just an awkward phase. We’re over it now.

Indeed.

The developer community deserves an apology from the ARB WG for their utter failure in communicating their internal problems. If the community had been kept updated on their progress or lack of it, the current backlash could have been avoided.

Of course, GLFW will still be adding support for WGL_ARB_create_context as soon as possible.

PS: Oh, and I saved up for and got a laptop with an Intel X3100, which means I can work mostly unhindered on GLFW again. Yay.

GLFW releases delayed

2008-01-13T00:10:00.000+01:00

Unfortunately, GLFW versions 2.6.1 and 3.0 will both likely be severely delayed, as Zepto Computers, the manufacturer of my laptop (a 6214W) refuses to pay for the repair and is asking for about $800 to complete it. As a recipient of disability pension, I don't even have $80 to spare, so even if I wanted to let them get away with this I still couldn't pay for it.

Their claim is that I should have blown compressed air into my laptop more often, and that the damage to the GPU due to supposedly clogged up fans (which were so enclosed that I could neither see nor access them) is my fault. Of course, they never mentioned this bizarre requirement in either their advertisements or their user's manual and the laptop was prone to GPU overheating even when it was brand new. With a GeForce Go 7600 with 512 MB VRAM in a smallish laptop, this is perhaps not so odd, but it's still their responsibility to make sure the unit is properly cooled.

Needless to say, I'm going to fight them over this, but it's going to take a bit of time, and in the meantime all I have is a five year old iBook with far too little RAM.

Also, if you have a Zepto 6014W/6214W/6314W and have had or are having similar problems, please contact me.

Times change

2007-12-08T07:07:00.001+01:00

I need to upgrade the software on my N73 and for that I need a Windows machine. As my own PC is currently broken, I was trying to think of a friend close by who might have one.

Okay, so here's the interesting part; it took quite a while for me to think of anyone. Note that this isn't "someone who's primarily running Windows", it's "who might still have Windows on their machine at all".

I think that means something.

What do you think?

Nostalgia, sort of

2007-10-30T22:23:00.000+01:00

So, I'm back on Mac OS X. Here's why.

The GPU on Sherka, my primary workstation, finally died. It did so about thirty seconds after I had completed a full backup of my entire home directory to an external drive. So I switch back to Sharon, my iBook, only to find that I've messed up both my Linux kernels' yaboot configs and the combo drive has died so I can't make a live CD. I'd be surprised, except that cluster coincidences like this one are far too common when you're an elmindreda.

So I'm stuck in Mac OS X Tiger on a five year old iBook.

However, I'm finding that it's not a bad place to be. A few apps later and it's a fully usable desktop. It feels almost like home. It now seems even more likely that the next laptop will be a MacBook Pro.

TPOLM Lazy Sunday Radio

2007-10-07T13:57:00.000+02:00

The Planet of Leather Moomins has struck again. Get your weird music fix here. It'll be playing most of the day.

Quickie: Super Mario Långben Bros.

2007-10-05T05:34:00.000+02:00

Someone made a SMLB film.

Hahahahahahahahahahahahahahahahahahahahaha <3 <3 <3 <3 <3 <3

Net neutrality exemplified

2007-09-23T19:05:00.000+02:00

Here is an image that in a simple and direct way explains why net neutrality is important. It should be made into a poster.

Using Fairly, in summary

2007-09-22T20:48:00.000+02:00

Thanks to Copyriot, I attended a (mostly) very interesting conference on copyright and creativity yesterday. Here are my summary notes from the many presentations given:

Being surrounded by and listening primarily to people who talk about art and the creative process was an interesting experience, however, since I've spent most of my life doing stuff and (compared to most people I know) not a lot of time talking about other people's stuff. This meta-world is quite new to me. It was nice, but I don't think I'd want to stay.

Robert Jordan, vanire ortekerana, donye aniar

2007-09-18T03:13:00.000+02:00

"The Light shine on you and the Creator shelter you. The last embrace of the Mother welcome you home."

Leyanade nay imino.

Akonade nay onaniar, vendesma so yasterana.

Kite lendiar? Robert Jordan passed away earlier today.

Quickie: So you want to be a game developer?

2007-09-13T19:21:00.000+02:00

This strip over on HijiNKS ENSUE is a good summary of the fantasy vs. reality of game development (possibly unless you're at id). Enjoy having any remaining illusions of the glamour of game development shattered. But it's still kind of neat to see your game on the shelves of... well, everywhere.

OLPC @ Google

2007-08-10T05:16:00.000+02:00

You think the OLPC is a toy for poor kids? You lose. Check this out. The Free Software world was given a chance to rethink the desktop, and it did. Now go dream about this technology (hard as well as soft) merging with next-generation phones, Internet tablets and media players.

PP @ Google

2007-08-08T17:55:00.000+02:00

Rick Falkvinge of the Swedish Pirate Party gave a talk at Google recently. If you want to explain (or learn about) Pirate Party politics, this is a good introduction.

"This time, we win"

2007-07-27T04:15:00.000+02:00

Seeing this site, I am reminded of nothing so much as Eben Moglen stating "This time, we win".

There are, of course, other signs. Nowadays, it is unusual for more than a week to pass without at least one significant piece of news illustrating this trend. There is, nontheless, something especially symbolic about this one.

Finger candy

2007-06-23T20:00:00.000+02:00

My dream came true! Well, one of them, at least. Thanks to NewsForge, I found Vimperator ~~today~~ yesterday, and oh my it's sweet. It's an Iceweasel extension that provides a very Vim-like interface. It has a few rough edges, but is despite those a vast usability improvement... for me. I suppose it makes my Iceweasel unusable for pretty much everyone else (most people I know use Emacs or something else lacking a command mode), but that's a price I'm willing to let them pay. My browser is now almost completely keyboardable, and that in a manner that's often far quicker than using the mouse. If you're a Vim user, you really should check it out.

The most important missing feature became painfully obvious once I started writing this entry; Vim keybindings inside text fields. They're still not modal, which now that the rest of the UI is feels even more weird than before. I wish I knew more about the internal structure of Iceweasel, because I have no idea how difficult it would be do implement this, but I hope it will be one day.

B

2007-06-11T02:58:00.000+02:00

I'm going to start writing about my little game project. As I've proven unable to keep quiet about it, I might as well use that to my advantage. Therefore, please comment, critique, suggest or complain about anything you feel isn't the way you'd like it to be or feel it ought to be.

Please note that I'm not making any promises, except that I will be taking all your feedback into consideration. If I like something you've written and it ends up affecting the game, you end up in the credits. Being an indie developer, that's the best I have to offer.

Also keep in mind that this is still rather early in the project, so don't expect fancy screenshots just yet. Just how early it still is will become clear in the coming months, and seeing as I'd rather over-deliver than over-promise, I'll try to puncture any expectations that might have arisen so far ^^

The completely uninspired working name for the game is Pod, which is now also a tag on this blog. Using this tag will give you quick access to all the relevant entries from now on. Just to get started, I'll post an entry I made to the internal milestone blog this past Saturday. This one is rather technical, but I'll be going into the game design and other topics in future entries, and possibly also posting additional internal entries.

Here goes:

I'm sitting at Compusphere listening to the streaming music compo and working on the core network code, built on top of enet. It's finally starting to turn into something implementable. Right now I'm trying to figure out where in the engine to put endianness conversion routines. Not the most fun you could have.

The goal for this milestone was to make something minimally demoable (to peers, not publishers) and I've reached the (possibly obvious) conclusion that a strictly multiplayer game can't be demoable in singleplayer. Therefore, I'm tearing apart my singleplayer demo and putting the network code in now. This means that all existing code (not that there's that much of it) will need to be moved onto the distributed event system, which will add further to the delay. Un-yay.

On the technical side, apart from the networking, I've replaced some hard coded stuff with actual level loading, split the game data into some sort of (hopefully usable) structure and started to integrate ODE via the Wendy wrapper interfaces. That last point has proven the trickiest by far and I finally understand the complaints of people integrating physics engines into games. Moving the game code to the latest version of the renderer was a very tasty piece of cake, however, so I guess the interfaces are starting to stabilise.

Design wise, lots has happened. The basic technology of the world is established, helping to explain why the gameplay looks the way it does and why certain things work and others don't. I've designed a few of the larger ships (ones players won't be flying, but will be fighting or be dependent on). I've also started sketching on the backstory. While it's technically unnecessary, it's something I really want to have, as I know it'll affect the design in interesting ways. I'm also thinking about ways to integrate a storyline into the game, despite it being exclusively multiplayer, but more on that later.

PS: The title is a reference to this entry. Planet readers should note that since the aggregator software is broken, it'll show up as new despite being six months old.

Brushed Metal is dead

2007-06-11T02:08:00.001+02:00

...and not a day too soon, either. Die the death, horrible monstrosity.

Here is how it happened.

Or not.

Lecture rambling

2007-05-24T12:22:00.000+02:00

I'm currently finishing up my preparations for an introductory lecture on rendering engine design, which is a follow-up to one I had a while back on OpenGL. Surprisingly, it's actually been trickier to prepare for this one, despite the fact that I supposedly know more about this subject. However, while the previous one was about a well-defined API and its mathematical underpinnings, this one is about best practices, common design patterns and other even more intangible knowledge gathered from experience. Success and failure aren't objectively measurable in the same ways here.

This has made it rather tricky to prepare the slides for the lecture, something I otherwise find both fun and easy. There's just no clear, linear path available to take me through a substantial amount of the area to be covered. In fact, the material is so self-referential that I'm finding myself wishing for goto in Impress.

What I will try to do is steer it towards more of a discussion between me and the audience, rather than just having me stand there presenting my preferences. I'm hoping that a few other sceners will be attending, which should help to make the discussion more nuanced (but hopefully not too advanced for the people who came for the promised introduction).

Oh well, I'm off to the campus to finish my slides.

I hate blogging memes

2007-05-22T10:42:00.000+02:00

...but this one was apparently irresistible:

B6 d- t- k s u-- f- i o++ x e- l- c-

GLFW braindump #3

2007-05-10T09:33:00.000+02:00

There, the first (and hopefully last) beta of version 2.6 is out and the amount of work remaining before the actual release is approaching manageable proportions. It's basically a few bug fixes, about 100 SLOC of Win32 FSAA code and a few paragraphs of text in the User's Guide.

Yay, I think.

I also note that the Win32 binary package is almost as popular a download as the *nix flavoured source packages, which says something about the user base. I hope I did a good job of building it, because a lot of the support requests lately have come from Win32 users unable to build the library (as might be expected from a platform with no standardised structures for development environments).

We've also have a web designer join the project. While he's busy right now, he'll be designing a new site for us when time and inspiration coincides. I have little idea when that might be, but I have every confidence that the result will be excellent. Having a less painful design to work against will be excellent, especially as I've been meaning to add a patch submitter's guide for a while now.

The discussions on the feature set and API for 3.0 have begun (tentatively) and as usual the timing is perfectly horrible. I need to spend the coming weeks working on the Birdie demo and the week after that finishing the material for my rendering engine design lecture. Then I'll be mostly free to work on GLFW, but that's far too long from now.

Update: Updated.

OpenGL on Vista; problem solved

2007-05-10T09:20:00.000+02:00

Just a quick note; many people wrote about Microsoft's plans of crippling OpenGL on Windows Vista, but very few have written about the fact that the issue was officially resolved last year (with an announcement to that effect at a BOF at SIGGRAPH).

Thus, the ARB (or what used to be the ARB) has therefore published a statement clarifying the current situation (i.e. OpenGL on Vista works the way it's supposed to), and I thought I'd help by pointing at it.

Reality wins again

2007-05-02T14:57:00.000+02:00

'38 214 62 45 178 91 204 116 247 110 121 234 76 121 167 239'.split.each { |s| printf '%02x ', s.to_i ^ 47 }

I'd like to thank the MPAA for playing. Good luck next time. We'll be waiting for you.

Update: Made it shorter.

Editor sneak peek

2007-04-24T21:21:00.000+02:00

I probably shouldn't be showing this before the Birdie demo compo, but I can't wait. Besides, I've already shown the effect visible in the render window (it's in my gallery on scene.se). The demo effect sequencer I've been wanting to make for several years now is finally a reality. I have three year old sketches of it in the official HyperCube sketch pad (yes, there is one), and the end result is surprisingly similar to those old scribbles.

Basically, you write effects in C++, register factory classes for them in the engine and then bring up the editor. This lets you create instances of the coded effects, place them in time, move them between layers and edit the properties they expose. All this is then saved to (and loaded from) a single, simple XML file.

It certainly isn't .werkzeug, but it does make editing and experimentation much faster, simpler and even fun. It's also gratifying that a few other groups are going the same route, and even more so that I seem to be ahead...

Oh, and before you complain about the horrid UI rendering, I do know how awful it looks and I'll be doing something about it as soon as the party season is over. Also, please don't ask why it has a Maali button, as I really have no idea anymore.

There, brag session over, going back to coding.

Uhm...

2007-04-24T20:35:00.000+02:00

Just in case you think your artwork is weird; it probably isn't. This, however, really is weird.

Lovely. Just lovely.

Please don't tell me what it's about. Even partial comprehension would ruin the complete awe I'm currently experiencing.

GLFW braindump #2

2007-03-15T03:42:00.001+01:00

I've been working almost exclusively on GLFW during the past few days, an endaevour much simplified by me getting a new keyboard for my laptop. That's one more point for Zepto, as not many manufacturers willingly send out replacement parts to customers and trust them to make the repair themselves. A few minutes and my lovely 6214W is once again the best laptop ever.

Uhm, yes, GLFW.

RandR support is nearly done, with just a few details left. Fullscreen on X11 now Just Works. I've begun to write changelogs and update the LaTeX documentation. The code for FSAA on Win32 is almost merged, with only the really nasty bit inside _glfwPlatformOpenWindow remaining. There's a bug in the FSAA support for X11, but I'm going to look into it once I've posted this, and it's on open drivers so I'm confident that it's our fault.

Looking at the changelog for 2.6, it's apparent that a great deal of work has been done, even though it's taken longer than I would have liked. I've had to learn quite a lot of new things in order to work effectively as project leader and primary maintainer of GLFW, from Xlib to diplomacy. I've also been really bad at keeping the community updated of our progress in the last few month. I really must do better at that, and if any of you read this, I apologise for the lack of updates.

I'm starting to feel comfortable as maintainer, however, and it's no longer daunting to look through the 20,000 lines of C that I was put in charge of (especially since I threw out or rewrote the really nasty bits). I still don't have a deep understanding of every line, but I'm now familiar enough with the code to make large changes relatively unhindered.

Once 2.6 is out, which shouldn't be long now, I'm going to rip out and replace some internal interfaces. No, it won't be an uncompilable mess for three months (and some of this work is already in 2.6, so I'm pretty sure of that), but it will make future development easier (and in some cases, possible). Both multi-window and multi-monitor support requires certain things to be changed, and I think both can be added for 3.0.

Successful failure

2007-03-12T01:13:00.000+01:00

Well, I set out to write a fastcompo demo and I failed.. However, I failed by producing so much material that I couldn't bring myself to cut it down to the recommended 30 seconds, so I decided to make it a proper demo. I can't afford to attend Breakpoint, and not being inclined to cheat I'll most likely release it at Birdie.

I seem to have hit some sort of sweet spot. Even though I'm slow and meticulous, even when I shouldn't be, I could code up effects almost as quickly as I thought of them. The engine (mostly) Just Worked, and the Lego feeling never went away. I even seem to have (mostly) avoided the post compo compulsion to rip the engine apart for major restructuring. There's just no real need for such things. Most of what I've been doing in the past week has been bug fixing and additional debug facilities. So, in summary, what I mean is this:

Min 3D-motor fungerar. Ellinor.

No, you're not expected to understand that.