Mesa Will Not Be Dropping Its Older GPU Drivers

[schmidtbag]

I know how to act depending on situation, and in forums this is the best way to get people to answer with useful information, or stop posting bullshit.

No, you don't know how to act. If you didn't behave the way I described, maybe the information you want to get across would actually get somewhere. Maybe people would respect your thought process more. Again, you're often not wrong, but your approach is.
It's basic human psychology - you aren't going to convince anybody you're right when you're attacking or belittling them. People don't tend to respond to you kindly all that often, nor will they admit you helped change their thoughts.
I am not claiming I'm better, but at least I'm actively aware of what's counter-productive. In the case of my original comment, I was venting; I had no expectation to be changing anybody's mind. After all, this is an informal comments section that mesa devs likely aren't reading.

I have nothing to lose if you hate me, and all to lose if you don't disclose useful info (and people posting bullshit piss me off).
So no it's not hypocritical.

I don't hate you, I hate your needless antagonism toward everyone. Also I'm not sure how others disclosing useful info has anything to do with your credibility (or whatever it is you deem lose-able). Regardless, what you deem "bullshit" is part of your problem - your arrogance. Even when it comes to strictly opinions, you will go out of your way to state your opinion as though it is "the truth". It's normal to disagree with people's opinions, and often people's thoughts are very mis-informed or incomplete, but how have you not realized by now that your approach isn't working?

I just pointed out that going through might have rustled other developer's jimmies (I've seen Marek that was against this in the mailing list, among others), and that in most cases where there is no clear leadership pissing off other people isn't good on the long run.

And since when has humanity ever come to a universal consensus? If change didn't involve naysayers, there would be no point in casting votes - we'd just make the changes no matter what.

I assume that the weight of keeping there the old crap is less than what we might think

Yes, that's probably true. But it's not just the bytes, it's keeping tabs on regressions too - that becomes a burden.

[mulenmar]

Why does it have to be this complicated?

Because humans are social animals who put interpersonal politics over getting the job done the best way possible, including those who act anti-social by framing everything as insults to make others look stupid in an effort to improve their own social standing. Also includes those who act all snooty and above-it-all commenting on said tendencies.

Just make a mesa-legacy branch and dump all the obsolete shit in there, If open source communities wouldn't let themselves be chained down by legacy shit all the time, focusing on progress instead of dwelling on the past (honestly it feels like some political correctness bullshit, don't wanna offend the old codgers who never bothered to upgrade from unicore machines, etc) we'd be so many miles ahead of everything else instead of just hanging on to Windows by the tail.

Yeah, basically what nVidia does with their legacy branch hardware support. Feel free to pay someone to do that.

As for supporting older hardware, some folks *cannot* afford to constantly upgrade their hardware -- they have to pay rent, overbloated medical bills (something something 'Murica), and don't want to go into debt. Others don't want to put working hardware in a dump to poison the groundwater and people around, or send it off to be "recycled" by dissolving it in acid in some African nation's backyard to extract the minute amounts of gold (and then burning whatever else is left), poisoning the land, air and people living around it.

Some have nostalgia from the experiences they've had with a particular machine, or maybe the modern stuff just doesn't meet the reliability or security (hello Intel Management Engine, AMD PSP, and firmware blob-ridden peripherals!) requirements. Some, in the laptop space, prefer the older form-factors of thicker laptops that had a decent number of ports, replaceable memory/storage, matte screens, and reliable cooling without costing over US$1,000 or worse.

Despite all that, I still have to tentatively agree with you that if someone wants to keep it working, they should put in the work to do so.

Dropping old shit from current software does not mean dropping support for the old shit anyways, the old software is still there and it still works, if problems arise then someone whom it bothers will pick up that torch. We need more trailblazers, the only reason why Microsoft Windows has been able to keep up with linux is because they're constantly dropping support for shit that would get in the way of progress. If we'd do the same by way of separating legacy from current (like how example Nvidia do with their proprietary drivers) we'd be making use of the overwhelming advantage we have instead of wasting it on bygones.

Microsoft? Constantly dropping support for shit?! HAHAHAHA, they still supported certain Win16 and old DirectX APIs, and they make sure as many ancient programs still run as they could, right up through Windows 7. Might've require installing some older Visual C++ library, or using one of the compatibility modes, but it was easily done, unless it was an old 16-bit program on a 64-bit version of Windows. They probably STILL support them in 8+, haven't tried and don't intend to.

[mulenmar]

Because it isn't what is happening already. When was the last time that someone actually used i915 for 3D anyway?

In my case, last week. 15" Thinkpad T60 with a good ol' 5:4 IPS screen is still good enough for many things, like running gzDoom at near-max settings, and parts are relatively plentiful. And, for everything else, just network to a more powerful machine.

[dungeon]

Guys, noone said anything about "deleting" or even "dropping support". Just moving them to another place, so they're not in the way of new features.

Neither AMD's for proprietary drivers ever said that they dropped support for anything. These are also just moved away to another place so they're not in the way of new features.

[brent]

As for supporting older hardware, some folks *cannot* afford to constantly upgrade their hardware -- they have to pay rent, overbloated medical bills (something something 'Murica), and don't want to go into debt. Others don't want to put working hardware in a dump to poison the groundwater and people around, or send it off to be "recycled" by dissolving it in acid in some African nation's backyard to extract the minute amounts of gold (and then burning whatever else is left), poisoning the land, air and people living around it.

The discussion is about branching out support for *really old* GPUs, around 10 years old and sometimes older. That is a far cry from "constant upgrading". These kinds of GPUs have limited use nowadays anyway. There's some discussion on mesa-dev about just enabling llvmpipe instead of using HW acceleration. And again, branching out into a legacy branch doesn't mean support will be removed. Users will still continue to be able to use the hardware (with less chance of regressions, actually).

[schmidtbag]

There's some discussion on mesa-dev about just enabling llvmpipe instead of using HW acceleration.

For clarification, is that a suggested solution as a way to replace these older drivers? Because if so that doesn't make sense - just about everyone using one of these older GPUs is also using a CPU just as old; a CPU that likely struggles to keep up even with proper drivers. For many of these old systems, they'd be pretty much unusable with llvmpipe.

[dungeon]

The discussion is about branching out support for *really old* GPUs, around 10 years old and sometimes older. That is a far cry from "constant upgrading". These kinds of GPUs have limited use nowadays anyway. There's some discussion on mesa-dev about just enabling llvmpipe instead of using HW acceleration. And again, branching out into a legacy branch doesn't mean support will be removed. Users will still continue to be able to use the hardware (with less chance of regressions, actually).

But these fixed function hardware like r100/r200 from classic mesa does not require llvm at all, also users or these likely has slow and probably mostly just single core CPUs... basically that helps nothing, with that you just wanna add dependency to provide nothing

[brent]

For clarification, is that a suggested solution as a way to replace these older drivers? Because if so that doesn't make sense - just about everyone using one of these older GPUs is also using a CPU just as old; a CPU that likely struggles to keep up even with proper drivers. For many of these old systems, they'd be pretty much unusable with llvmpipe.

Slow software rendering is still better than a broken driver that does not work at all.

[starshipeleven]

If you didn't behave the way I described, maybe the information you want to get across would actually get somewhere.

My information usually gets across where it is actually possible (i.e. the person is listening and able to change his mind). Where it doesn't work also other approaches would have failed. Anyway getting info across isn't main goal, I share info where possible, but it's not a goal.

It's basic human psychology - you aren't going to convince anybody you're right when you're attacking or belittling them. People don't tend to respond to you kindly all that often, nor will they admit you helped change their thoughts.

As said above, my goals are either data mining or stfu-ing people I think are posting bs, so far I'm reasonably successful at both, not just here.

I know enough about human psychology that convincing people is a major PITA to do when in person (go figure over a text-only media), and some beliefs just cannot be changed, period. So I don't aim for it, at least not directly. You want a good read? Try reading How to Win Friends and Influence People by Carnegie. It's kinda ancient but is 100% fine as men didn't change.

Regardless, what you deem "bullshit" is part of your problem - your arrogance. Even when it comes to strictly opinions, you will go out of your way to state your opinion as though it is "the truth". It's normal to disagree with people's opinions, and often people's thoughts are very mis-informed or incomplete, but how have you not realized by now that your approach isn't working?

Doing that both opens the flank to anyone that might actually know more so he comes and corrects me (data mining) or has the effect of getting idiots to stfu and leave eventually.

Sorry but I tried also other more "civilized" methods, and they didn't work. People are more likely to come and answer if you offer them a bait (i.e. proving an arrogant guy wrong) than if you don't.

And since when has humanity ever come to a universal consensus? If change didn't involve naysayers, there would be no point in casting votes - we'd just make the changes no matter what.

Since you clearly didn't understand the point, pissing off your collegues over stuff that isn't major isn't a good way to go forward. If you push and piss off people it better be worth it. It seems he thinks it isn't.

This isn't the first time we talk about group work and how your very rational and machine-like views clash with the actual reality of many workplaces.

[starshipeleven]

In my case, last week. 15" Thinkpad T60 with a good ol' 5:4 IPS screen is still good enough for many things, like running gzDoom at near-max settings, and parts are relatively plentiful. And, for everything else, just network to a more powerful machine.

Damn I hate you for pissing on my parade. Stop immediately, I dare you.