Log in

No account? Create an account
Further deceipt from intel's Keith Packard. - LIBV Intentionally Breaks Videodrivers [entries|archive|friends|userinfo]
Luc Verhaegen

[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

Further deceipt from intel's Keith Packard. [Jan. 21st, 2007|04:13 am]
Luc Verhaegen
[Tags|, , ]
[mood |annoyedannoyed]
[music |The Pixies - Surfer Rosa - Gigantic]

I just watched Keith's talk in the debian miniconf room at LCA.

My. My. Randr really is the best thing since boiled water, isn't it.

Now, a lot of what he says quite categorically hasn't come from himself or Intel, and only manners stopped Keith from calling it all "Rubbish!" at EXDC. There's also the fact that no-one currently working on the intel driver ever bothered at all with anything modesetting before March 2006. That was all established in my previous post. So most of what Keith says has been rehashed from, to a sizable extent, my work and views. This in itself isn't wrong, it's the run up to it, and how it all is happening that's completely wrong, and of course the fact that Keith doesn't at all want to acknowledge my and other's part in all of this.

Least of all admit the gross error he made at EXDC. If anything, Keith slowed down proper modesetting by a year there.

Anyway, have you taken a good look at the slide with the feature set for Xorg 7.2? If not, look again.

Do you see anything missing there? Maybe not, but I do.

7.2 is able to run without a conf file. It's not perfect, but it is work-able. You'd think that this is a nice feature, you'd think that it would be worth mentioning, wouldn't you.

So why is that missing from that slide? Why does keith only mention it in passing much later?

Most of the work there was done by Adam Jackson, but most of the modesetting work needed there was done by yours truly (part of it was also done by Adam, but that move was driven by redhat corporate profiling/short term views versus bedroom hacker/slightly longer term views, and this is another story altogether). These parts are CVT modeline generation (in server too), and actual handling of EDID data (not parsing, that was done last millenium by Egbert Eich).

Sure, these bits aren't rocket science, they are just direct implementations of known/documented things, but they are absolutely fundamental to everything modesetting.

So fundamental to modesetting, that intel was unable to go around it. They now even directly include it in their code.

But this is still no explanation for not mentioning it in this talk, but i do have some hypotheses.

Mentioning this would:
* reduce the deceptive novelty of randr1.2. Decrease the number of Oohs and Aahs that are so essential to an enormous ego.
* point to me, and my views, ideas and code, and would expose keith and intel, and what's behind this whole story.

Whatever mix of these it is, it should be clear by now that this is all very deliberate. Add it all up, and it clearrly cannot be just a collection of honest mistakes.

This is a tactic.

This is about:
* pure unbridled marketing.
* personal scoring, no matter what the costs are for X.org in general.
* chronic Not Invented Here.
* money.
And i'm certain that a lot of it is/was quite personal too.

Sure, you might not believe all of what I am saying, but there is quite a lot of hard evidence. And even if you only believe half of it, you're still seeing the actions of a person who's not exactly worthy of the title of open source hero.

These are the actions of a monopolist. You are seeing a open source intel monopoly in the making, because these are nothing but monopolist tactics.

Keith Packard _is_ intel. They were made for eachother.

Here Be Dragons.

[User Picture]From: korgmeister
2007-01-21 12:00 pm (UTC)

It'd be nice if I could talk about rubbish drivers without swearing constantly. But I can't.

7.2 is able to run without a conf file. It's not perfect, but it is work-able. You'd think that this is a nice feature, you'd think that it would be worth mentioning, wouldn't you.


The reason I ended up switching away from Linux back to Windows was because one time, when I did an apt-get update on a sid install, some shit-for-brains at some point in the process (was it Xorg, was it the guy who coded the driver? Was it the fucktards in Debian package management? I'll never know) thought it would be really clever to start changing variable names in the .conf file in a seemingly arbitrary fashion.

At least, I think that's what happened. All I know is that the scripts which were supposed to write a whole new .conf file had not Got With The Program because they didn't fucking work. Oh and the really fun part was that it would randomly work about 10% of the time so it was damn near impossible to diagnose whether my actions were having any positive effect or not.

So yeah, anything which potentially stops someone else from having to go through that CLUSTERFUCK is an Incredibly Good Thing.
(Reply) (Thread)
From: (Anonymous)
2007-01-21 08:44 pm (UTC)

Not the right approach

I'm just a user. I don't know either of you guys, or the situation in detail. But, regardless of what is going on, I don't know if this kind of posting really helps anything. I think rather than blogging negatively about others, you would be far better off to spend that time blogging about the work you are doing in X, or about things going on in X which interest you. If you think people are getting a skewed viewpoint on how things are going, blog an alternate viewpoint saying "I think XXX is a good/better idea because...", rather than just bashing things/people. I read this blog on Planet Free Desktop, so, from my POV, you have just as much as a voice in the community as Keith.
(Reply) (Thread)
From: (Anonymous)
2007-01-22 12:28 am (UTC)

Re: Not the right approach

Let me echo that. x.org is a very complex piece of software. Kieth does a lot of useful work. He is wrong sometimes no doubt, but so is everyone.

Post on what you are doing. Post a vision. Do the work to make things happen. Sometimes someone will take credit for your work, but if you keep doing useful work they will not be able to continue.

It is much better to have others defend you, than doing it yourself. You are too biased. Already it appears you cannot see the useful work Kieth is doing.
(Reply) (Parent) (Thread)
From: (Anonymous)
2007-01-22 04:08 am (UTC)

Re: Not the right approach

Yeah, I agree. Keith has done a lot of good things. He might have messed up here. That happens. I don't really feel like he's somebody who has bad intentions. After all he's done and all I've read from him, I don't see him as somebody who is going to really screw other people...and X...over.

I would like to see Keith respond to this.
(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-23 01:37 am (UTC)

Re: Not the right approach

Well, i'm not just bashing people. I got bashed. I got bashed and then my views and ideas were very directly taken and said to be the greatest thing ever. And the exact same person did both, this while this same person does absolulety everything to minimize my part in it all: I never did exist.

Why? Because Keith Packard is utterly unwilling to admit that he was completely wrong.

Now, a lot of what was described was before i started blogging (which was august 2005, 2 months after EXDC), and i never have used my blog to talk about code much, i used it more as a place to vent. Code happens mostly in revision control systems, on bugzillas, on mailinglists and irc. I could've also whined more extensively about modesetting here, but things were boring enough here already.

I wonder why you think that i am so biased (of course i'm somewhat biased): how much of what is said here is in doubt?

On the other hand, i'm not at all wondering why there's no-one else defending me. Exposing Keith Packard is political suicide, and i waited very long to do so because of that. Even I realize some things about politics. Now, everyone who would like to think that he is part of some X.org community and would like to uphold that illusion for a while longer will not move a muscle. Some of those so unbiased folk are comparing me to David Dawes already. This solely on the basis of being critical of Keithp i think.

No miracle that keith does useful work. Intel is paying 12 people directly for this stuff, and some are getting paid quite a lot more than others. Imagine him _not_ doing anything useful. Part of why he's laying things on this thick now is because he really must make out as if he is doing useful stuff, while i'm very certain that he at least initially dreaded modesetting completely.

He waited for a long time to take the output route, and tried to stay as far away from what I've been preaching and doing for as long as possible too. Output "properties" are quite telling in that respect: "Oh, bugger, a tv-encoder. It seems like libv knew what he was talking about again, but let's never ever admit to that."

This is not a simple mess-up. This was a serious mess-up initially, then a cover-up, with a lot of spin and a lot of things being "accidentally" overlooked.

He is screwing people over, and no-one can still deny that the recent part is deliberate.
(Reply) (Parent) (Thread)
[User Picture]From: airlied
2007-01-23 09:25 am (UTC)

Re: Not the right approach

Look Luc no-one in X.org is joining in becasue you are just bashing ppl who earn money working on parts of X that you happen to have worked on, I'm sorry if working as an independent developer isn't bringing you the riches you wish for, but funnily enough I'm in a similiar state and I manage to get paid but I don't go around giving out about anyone and everyone and bashing people who have commerical schedules or pressures for stealing my ideas and livehood...

Ppl from X.org have watched you incessantly bash ajax and RH on irc for putting your code in FC6 before it got merged into master, even though you give your code out under a BSD license... this was mainly around the fact that RH paid ajax and you were getting nothing for your work...

I wrote some basic drivers to talk to i2c dvo devices, funny enough I didn't get it finished before Intel needed it, they took my code stuck it in their driver and finished it, you don't see me blogging that keithp/anholt stole my ideas...

So sorry you don't earn money from your X.org code, but if that was your primary reason for working on X.org you were in it for the wrong reasons...
(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-24 04:32 am (UTC)

Re: Not the right approach

Oh, sure, same wrong stick again. Don't you have anything else to say?

Anyway, let's start dispelling this first, before i tackle bug #5386 or other false truths in your comment.

I have never been in it for money, because if you're in it for money, you choose popular subjects, and popular hardware. Since when is modesetting popular, and since when i have been almost singularly doing modesetting? How popular is the VIA unichrome, or the tseng cards, or the VGA driver?

If i were in it for money, i would, i think, make sure that i'm present everywhere, on the most high profile topics and hardware. I'd be there when projects and ideas already have traction, and act as if i was part of it all for a long while. I'd quickly do some work to proof actual involvement and then hold talks about such projects, trying hard to run away with a lot of the credit.

Why is it, also, that you are always the only person claiming that i sollicit payment all the time?

Now, i have become self-employed, which is not something anyone does lightly here. I became self-employed for a single reason: as a framework to support the development i was doing. At unichrome.sf.net it was clear that:
* There was and still is a major shortage of hardware. If you do take drivers and modesetting serious (and not do, "look, dualhead, but only CRT, now i can act like i was the main developer all along"), you need hardware. And when you're doing IGP, you need to get a rudimentary equipped but full PC each time round.
* Many companies were using this code and some were looking for a way to support development and get their issues resolved.
* I was, at the time, totally unable to accept anything. It was that bad that a nasty Tax or un-employment officer could take my work on unichrome.sf.net to mistake it as part of the black economy.

So, the only option was to take the self-employment course, fill out tons of paperwork, and take a leap, a leap that's not at all straightforward in a country like belgium, where social security rules, but where you lose a lot of it when you become self-employed. This is a level of dedication (or folly), that you never had and never will have.

I have always firmly maintained that i will not take money from non-commercial users. Which leaves me wondering how that 10kusd nouveau donation fund will work out for you.

Now, except to commercial unichrome users, i have always solely been complaining about -ENOHW. Sure, this is about a lack of money, but have you ever heard me complain about not being able to buy food or clothes or whatever? This while i never did manage to pay for those things either, i never complained, that burden is being fully carried by my family.

Most of what i received (from family) has always gone to hardware, but that amount doesn't exactly run up to the 2-5K needed yearly to be able to stay in touch with VIAs IGP release cycle. And the tiny amount i managed to sell so far paid for 1 quarterly social security bill (60%), hardware (rest). You made more from your company when you spent the two weeks on the intel driver, which was something you would never have attempted in your free time.

Oh, and FOSDEM last year: every penny received from X.org went to food. Yet i only received money from X.org more than a month later. I had pretty much maxxed out both bankaccounts on thursday already, and i had to ask Matthias for his half of the hotelroom to be able to pay for the sandwiches on sunday. The one time i did get payment for, not from X.org, but from google, was linuxday dornbirn 2005, when i was desperately trying to save up to become self-empoyed. The price for Dornbirn was a working VIA EPIA SP, and I received a partial refund 6 months later, at which time i used it to pay for the trainticket to linuxtag.

Feel free to outline what you've received from fd.o or X.org, or even intel recently (why did you travel to the us last summer?), so far.

So, here you are. Continuously saying me that i'm in it for money, but you're telling more about yourself than you are saying about me.

I suggest you find another stick to bash me with, Mr. On-Keithp's-wagon man.
(Reply) (Parent) (Thread)
[User Picture]From: airlied
2007-01-24 04:56 am (UTC)

Re: Not the right approach

Have you asked why commerical unichrome users don't support you?

I've always blog about all h/w I receive and all trips I take and who funds them, funnily enough these are usually not much use to me financially or personally (if you enjoy doing 14hrs in a tin can each way for 2 day meetings then you are welcome to take them..), they are more to do with the fact that I am the DRM maintainer in the kernel and ppl want patches tested on their hw before release, you might note the drm before I got involved was an area which was very neglected and very very unpopular so I don't see how your comments above fit actual life..

I picked radeon because my customer picked radeon, I moved to Intel because my customer picked Intel, I worked on nouveau because I had a box with two monitors and no way to use them... as for nouveau I've already stated many times that I'm not interested in getting anything from the us$10k as I have no great want for nvidia hardware, I'd think about going to a developers meeting if funding was there but only if it was really necessary which I doubt it will prove to be.....

The reason I think you are in it for the money is that you constantly complain and give out when a funded developer works on anything you might have stated an interest in, from ajax and Redhat to keithp and Intel, this is just a reflection from your public statements on irc and blogs, maybe you have another issue that causes you to attack people who use your code under the license you release it under... I've even seen you go crazy on irc when ajax discovered that they had a monitor database and accused redhat of conspiring against your work by covering it up, despite the fact ajax had just himself discovered it..

As I said all trips and hw are available on my blog and always have been, X.org have funded one trip to XDC and Intel have funded one trip to a kernel developers meeting in my capacity as drm maintainer, in that case I didn't ask for it, they asked me... so yes keep up the mad rants and conspiracy theories but you really need to figure out how much of the problem is in your own head or move on to looking into JFK's shooting...

(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-24 09:56 am (UTC)

Re: Not the right approach

> Have you asked why commerical unichrome users don't support you?

* Small companies don't often intend to pay for things they assume they get gratis in the first place, so it is a very small market to begin with.
* My choice of modesetting versus bling, and this long before anyone currently talking about modesetting ever saw it as a worthwhile cause.
* Dwindling popularity of the VIA unichrome, mainly due to VIA being a clueless vendor and intels marketing blitz.
* My uncompromising personality: when i'm right, i do not give in, at all.

"anything you might have stated an interest in"

That's bending the truth quite severely. My work on modesetting is, ever so slightly, more than just "anything i might have stated an interest in".

Now, bug #5386 and some other things, that's for a seperate blog entry.
(Reply) (Parent) (Thread)
[User Picture]From: mjg59
2007-01-24 03:47 pm (UTC)

Re: Not the right approach

My choice of modesetting versus bling, and this long before anyone currently talking about modesetting ever saw it as a worthwhile cause.

It was clear that the Intel driver was going to have to gain modesetting support at some stage ever since vendors started shipping machines which didn't have the panel mode in the BIOS. Which was, ooh, about three years ago? Alan's work on the Intel driver was always constrained by the fact that Intel seemed to want him to do as much as possible through the BIOS, something that's only changed since they took driver development in-house. Saying that nobody cared about modesetting on Intel hardware before 2006 is just clearly wrong, especially since I had a discussion with Keith about what would be involved at LCA in 2005.
(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-25 02:59 pm (UTC)

Re: Not the right approach

It was clear that it was needed, but everyone was saying that it was utterly impossible to change. Except me, as i knew how i was going to deal with outputs.

And talk, there was plenty of talk all around. But there were only two people really involved in modesetting. The most experienced of those was Twini.

So now I'm left wondering what exactly was being discussed at LCA. What solution did keith see then?
(Reply) (Parent) (Thread)
From: fooishbar
2007-01-24 07:31 am (UTC)

Re: Not the right approach

This is a level of dedication (or folly), that you never had and never will have.

Having heard a story about changing into a suit for a job interview (without having showered) a few hours after having arrived in the country for good, with no place to live, I have to say you're wrong.

Are you going to apologise, or just keep on being an antagonistic twit?
(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-24 09:59 am (UTC)

Re: Not the right approach

Apologise for what?

For being an antagonistic twit? How about an apology for this serious and deliberate lack of correct acknowledgements for CRTC<->Output separation.

Oh, and since you're here, how about confirming the discussion at that EXDC lunch?
(Reply) (Parent) (Thread)
[User Picture]From: airlied
2007-01-24 10:35 pm (UTC)

Re: Not the right approach

Apologies for stating that "This is a level of dedication (or folly), that you never had and never will have." towards me for a start. I moved across the world with no job and by the time I got here very little money (which is what daniels pointed out...)

From my knowledge of keithp and this output/crtc stuff he probably didn't remember so much of what you told him as when we he moved to Intel and randr-1.2 was raised at DDC he didn't realise the CRTC/output split was necessary (I'm not sure if the initial draft of the randr 1.2 is available but it certainly didn't have the split.....), so I guess you are just attributing something to malice that is actually attributable to the fact that no-one listened to you, which you need to attribute to a bad signal-to-noise ratio....

(Reply) (Parent) (Thread)
From: fooishbar
2007-01-24 11:23 pm (UTC)

Re: Not the right approach

Yes, for being an antagonistic twit. I can't be bothered arguing with you about Keith, since you obviously have your heart set on that one, but you're certainly wrong about Dave, as I've just shown above. So you owe Dave an apology.

And I don't remember every conversation at every lunch I've ever had (did you disclose that Intel, by way of Waldo, bought lunch that day?). I vaguely recall someone mentioning that input badly needed fixing, but not any details. (I didn't start working on Xi stuff until we needed it at work.) So no, I can't confirm that.
(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-25 03:11 pm (UTC)

Re: Not the right approach

I'm not convinced about Dave, and as long as I'm not convinced, there's nothing to apologise for. Maybe in future, but certainly not today.

In fact, i should be getting apologies. My abstraction was severely bashed by Dave, and this was very much unjustified. Dave is the sole reason for seeing my copyright scrapped here. The subsequently built up abstraction, built up from the callbacks Dave kept, after having scrapped the rest, is so highly parallel that it's no coincidence.

Now, i find it rather convenient that you claim to have forgotten as much about that lunch. I'm certain that you know full well that Keith cracked down on the modesetting stuff. You might not have detailed knowledge anymore, as this wasn't such an important moment for you, but you will be remembering that much.

And the fact that Waldo did pay, after having stated "Intel will get this one", doesn't change anything about what was said.
(Reply) (Parent) (Thread)
[User Picture]From: airlied
2007-01-25 11:13 pm (UTC)

Re: Not the right approach

Hey I didn't steal you i2c drivers, I started that from scratch it looked nothing like your code, as you code was abstracted enough from the drivers for it to be easily useful for intel/radeon/nouveau, since then Eric Anholt as re-written the interface yet again to make more sense for Intel and I still may need to tweak it for radeon,

Yes I looked at your code for ideas, no I didn't steal it the code as I didn't agree with your design for the i2c drivers, which was all I cared about I didn't feel the need to re-write my driver from scratch just to use your i2c driver for one chip...
(Reply) (Parent) (Thread)
From: (Anonymous)
2007-01-27 11:11 pm (UTC)

Re: Not the right approach

You still don't get it, do you.

It wasn't about I2C drivers alone, it wasn't about modules alone. It was about providing a clean yet complete abstraction, which would handle both external as internal outputs, and which would also directly allow modularisation.

Yes, my initial mindset was "It's modular in hardware, it needs to be modular in software". But by the time you came round I had abstracted the CRT (or was this one a bit later?) and the (TTL) panel (this was abstracted quite early on) too, yet you still talk about I2C.

You still don't see it.

This was why you said i was all wrong, and why you only kept the callbacks. You did, quite unmistakedly, take those straight out of my code.

What did anholt do; well, he built up an abstraction again from those callbacks. And there's no denying that he spent a lot of time looking at what i was doing in the unichrome driver.
(Reply) (Parent) (Thread)
From: fooishbar
2007-01-26 02:22 am (UTC)

Re: Not the right approach

If you still think that you are fit to insult Dave because you quit your job and decided to start your own business while still living off your parents, while he moved around the planet with not much money and no job at all, then congratulations to you; your reality distortion field/ego is quite impressive.

Remember that everyone has equal access to the tree. Because you decided to rewrite Mach64 ISA probing or something before you finished your modesetting work, and then quit in a subsequent tantrum, is no-one else's fault. You had just as much right -- and time -- as ajax to merge the modesetting stuff. You'll note that ajax went to pains to preserve your name on all commits authored by you. How was this even remotely related to stripping attribution?

In fact, I don't even care. I'm telling you that I don't remember what went on at that lunch, because I don't really remember what went on at that lunch. Waldo was taking notes on what was important for Intel: it wasn't an exercise in my education. It was just another reasonably enjoyable chat. I've had many of those over the years, and I don't even remember half of the ones that took place (I'd forgotten about this one until you reminded me), let alone have taken notes.

You've got your own parallel reality where Thomas Hellström, Ivor Hewitt (apologies if the name is wrong) and the others are all impossible bastards who forked the .*chrome stuff for no good reason, and in which all of X.Org is impossible to work with. But everyone else is wrong, damnit, and I'm the only sane person here. I know someone else like that; maybe you should get in touch with David.
(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-27 11:56 pm (UTC)

Re: Not the right approach

It's amazing how, in that distorted reality, the same thing always happened:
insights and ideas got:
* at first bashed.
* then ignored.
* suddenly, 6 to 18months later, rehashed and deemed the greatest thing on the planet.

Maybe, just maybe, i look slightly further than right-now, and maybe, just maybe, only right-now and "what my manager wants" is the only truth you and most other X developers chose to accept.

Anyone with a longer term view, without corporate credentials, is dangerous, it seems. And he should therefor be stopped at all costs.

Iirc, the mach32 was available as both isa and pci, mach64 was pci-only.

The modesetting work (which was, first and foremost, cleaning up xf86Mode and turning xf86validateModes into a compatibility wrapper around many helper functions.) meant passing through atimisc and significantly reworking it's modesetting code. Since atimisc probing was also stopping the libpciaccess work from becoming fully functional, i chose to also tackle that, especially since i had many months until the next X release. 7.1 was faar too optimistic back in late february, 7.2 was at least 8months away, which became 10+.

But ajax wasn't on an X release shedule, but on a redhat release shedule. Look at how timely 7.2 was. He had also, clearly, made some promises to his manager about EDID handling for FC6. Feel free to also notice (for a change) how he kept all his code tucked away in rawhide, and didn't bother to put any changes back where the code originally came from until he said "I think it is ready.", and even then it took many days for it to appear on git.fd.o.

I'm the uncooperative git, right?

The unichrome fork happened over VBE and the importance of modesetting, over unichrome naming, over the importance of hardware, and over the parasitic intent of some commercial users play(ed) and how to treat them, ... Package that up undiplomatically, and the fork is fully explained. On all these points, time has proven me to be completely correct.

Not that I've always been correct, there are many things i was wrong on. But on those things i really fought for, I can't remember a case where i wasn't proven correct by time. Yet I was always deemed completely wrong, yet afterwards all people (like you) care to remember is how completely wrong libv was on,... something... can't remember what anymore... But libv was wrong!

I really can't wait until tungsten lands a contract to deVBE and randerize xf86-video-via. That would, once more, be X.org perversity at its best.

I'm sure that the company coughing up for that would enjoy a bit of a history lesson.
(Reply) (Parent) (Thread)
[User Picture]From: airlied
2007-01-28 07:13 am (UTC)

Re: Not the right approach

How were you right with respect to the forks for *chrome?

As far as I can see the situation still one of almighty stupidity, you have your driver, openchrome as theirs and X.org and distributors ship whatever is in the X.org driver which I think Thomas is probably updating more often than anyone else...

So the users all lose, as they don't get out of the box support for their hardware and you sit around shouting at the sky that you were right all along and everyone else is wrong, as far as I can see you have proven nothing wrong other than to make VIA hardware less likely to work on users hardware running Linux, you obviously don't understand that most users never find your driver they run what comes with the distro and complain to the distro if it doesn't work... so I doubt if you count success as lack of bug reports it means much more/less than lack of users...

And there was ISA mach54 unfortunately for everyone, granted I think we can deprecate support for it at soon without many ppl noticing...
(Reply) (Parent) (Thread)
[User Picture]From: libv
2007-01-29 06:10 am (UTC)

Re: Not the right approach

I was 100% right on the directions i chose, and these choices caused the fork.

Then, i wanted to see xf86-video-via dropped altogether, as this only further confuses the situation. I was then told that what I was doing was a complete regression, and shouldn't be brought into Xorg releases. Add to that that my code is quite fundamentally and significantly rewritten, whereas openchrome has just gotten EXA slapped on.

Oh, and erm, I've always had to hear that the via driver must be good, because people like mharris never saw any bugreports for it, and that therefor i was such a bag of wind.

How you even dare to use "why don't you update xf86-video-via" is beyond me.

Well, atimisc is/was holding back a lot, you obviously will not want to admit to this, but you know it's true. I still need to see you even attempt to do such important clean-up work.

I've always maintained that X-wide ISA and VLB support should not be dropped altogether, but that we should be stripping drivers from such support if that support is holding back things. Atimisc is/was blocking a lot.
(Reply) (Parent) (Thread)
From: (Anonymous)
2007-02-17 05:47 pm (UTC)

Re: Not the right approach

"Then, i wanted to see xf86-video-via dropped altogether,"

...and exactly how the fuck does that help anyone?
Who exactly are you in this for?
(Reply) (Parent) (Thread)
From: ext_29620
2007-01-24 06:35 am (UTC)

Running without xorg.conf is not new

I wouldn't list running without a config file as a new feature of Xorg 7.2 - I've run without one since the first Xorg release (6.7), and since we first shipped Xorg 6.7 for Solaris, we've shipped with no xorg.conf by default, letting people add one if they need it. If I remember correctly, this feature was first added by David Dawes for XFree86 4.3 (http://www.x-oz.com/autoconfig.html). The defaults selected without an xorg.conf have needed some work and have gotten better in each release - and the modesetting and other autoconfig work done in 7.2 is no exception, and included in lists I make of new features in 7.2, but xorg.conf-less operation is not new.
(Reply) (Thread)
From: live213
2009-02-09 04:39 pm (UTC)


(Reply) (Thread)
From: links12
2009-02-09 05:00 pm (UTC)


Never a thank you to write this is not spam I wrote with my own See You...:)
rent a car, araba kiralama, araç kiralama
(Reply) (Thread)
From: (Anonymous)
2009-06-04 05:45 pm (UTC)

Chuckle modesetting games in 2006...

ATI was already planning it in late 2005, releasing it in early 2006. Well and truly before it was formalized in RANDR 2007.


(Reply) (Thread)
[User Picture]From: libv
2009-06-05 10:15 pm (UTC)

Re: Chuckle modesetting games in 2006...

Yes, i was informed that ATI was looking for some sort of a solution in this direction. I wasn't working for SUSE yet at the time, but i'm guess that some my message somehow got through by that channel.

On the other hand... I did end up constructing a saner modesetting model for proper graphics when i did work for SUSE and developed the radeonhd driver base modesetting. This too good "BOO"-ed away, especially by those people implemented something highly similar in KMS a few days later. History repeating...

Now... I refer to an older blog entry in this entry, it is the one where i outline what happened in before RandR1.2 was proclaimed the one and only solution. It talks about the time when keithp still wasn't working for intel and he couldn't be bothered and he too was actively BOO-ing what later got reinvented (rather badly) and called RandR1.2. That story predates the ATI story quite a bit, it predates any modesetting story, as people were still actively trying to kill what got named xf86-video drivers after the modularization, and everyone proclaimed kdrive as the one and only solution (see a parallel with kms here?).

It is amazing what people do get away with, and how shamelessly they do so time and time again.
(Reply) (Parent) (Thread)