It took them a week since xorg-server-1.18 was tagged. But xorg-server has a public git repo, so nVidia could track its development as a matter of continuous integration, if they so choose. If resources are a problem, well, hey, they know exactly where and how they could get dozens of volunteers for free.

I assume it's because -current occasionally ships updates (including this, but I'm mainly talking about kernel updates) that require you to reinstall the NVidia drivers.

That does not, IMHO, mean it's "ill-advised".

1 members found this post helpful.

Take it however you want it, but the statement is what it means. -Current is always changing and something could break a proprietary driver package without warning, as we just saw. You know FULL WELL, and don't need to be told repeatedly that the nature of -Current is not well suited to proprietary drivers under any circumstance and using them carries some risk as to why I, and will repeat for yet another time, they are ill-advised to use.

Kernel updates, and it will break the driver.
Xorg-server updates and it will break the driver.
Mesa updates and it will break the driver.

Yes, getting the proprietary driver(s) tested is one thing, and THERE IS A THREAD FOR THIS THAT TARGETS SBO, but testing it against a stabilized release, and not a development branch that is always changing, and can promote breakage is not the wisest of ideas, nor is going to be able accuractely bring about a stable package when another kernel update, xorg-server update, or mesa update could come and foul everything up again.

Take what Patrick said to heart and hold it well. The man doesn't need to be quoted, but you'd damn well be wise to heed his words.

Please provide proof to this statement other than nonsensical hearsay.

1 members found this post helpful.

So, what's the problem? It is SOP that you should always re-install the driver after those packages are updated. You might add gcc and glibc to that list.

No scuttlebutt about it. Let us be realistic and not open source zealots, the Nvida driver has, overall, better performance. .

2 members found this post helpful.

SOP=Standard Operating Procedure? I have blacklisted kernel, mesa and xorg-server for that very reason.

And yet you guys complain about a proprietary driver that isn't properly supported by the current software packages...

Yeah.

If that is directed at me, no, I'm not complaining. I wouldn't run -current if I didn't expect there to be a problem every now and then. Fixing it is part of the "fun."

1 members found this post helpful.

Exactly, but please refrain from saying what is actually better, because that is a per-user case, and not an absolute fact, and also remember that an unstable development OS is not exactly the ideal platform for using proprietary drivers, much less debugging/testing them.

The free driver might not be the most comprehensive driver, but 9 times out of 10, it works.

1 members found this post helpful.

Yes, Standard Operating Procedure.
I do the upgrades and changes the "old fashion way," by hand, i.e., at the command line, and don't use slackpkg. BTW, I read somewhere, one should not upgradepkg the kernel, but installpkg them and edit lilo.conf accordingly. Good advice.

If you want to do research and post the results of any neutral testing of noveau vs. nvidia you can find, I'm sure many of us would be interested.

You brought it up, so that's on you sir.

Besides, my driver the 340.xxx series is actually worse than xf86-video-nouveau. While it is branded by Nvidia and blessed by their dark unholy demonically possessed sages who develop drivers from the putrid ether of the forever-dark netherworld of the nvidiaverse, a place even deemed off limits by Bob himself and even chastised by Linus Torvalds with a middle-finger, I now get better performance and support through Nouveau and I have a 9800 GT (which isn't that old mind you).

What are you talking about? This is the state of any 3rd-party program on -current. An update could knock everything out.

I would never say that -current users shouldn't use proprietary drivers, but, as with anything related to -current, things could break and you should have at least a basic understanding on what is going on.

By your standards, -current wouldn't be good for alternative inits, different WMs/DEs, etc...

Now, users who run current, shouldn't automatically assume everything is going to work perfectly. When -current is as bleeding edge as we are on some things (like mesa and X), breakage is very possible (especially in regards to nvidia drivers, since they seem to wait until bug reports come in rather than keeping up with development), but to say that proprietary drivers are not recommended is just absurd.

Now, if you wanted to throw a little addendum (my words in bold) on the end of that, it would make a lot more sense.

We should be finding these problems so things can get updated. If it isn't something in Slackware that needs to be updated, then it may lead to a SlackBuild being updated, or finding alternate versions, or even alternate programs (obviously not possible when dealing with proprietary drivers unless you switch to opensource). The Slackware forum doesn't only cover the development of official Slackware, but also covers people's use cases. Finding out what needs to be done to get the proprietary driver working is what (some) of us are here for.

2 members found this post helpful.

Here you go:

Linux kernel 4.1.12, GeForce 9800 GT. All I get then is the VESA driver when starting X. Would be nice if nouveau was working, but unfortunately it does not even initialise on my card.

1 members found this post helpful.

The last i heard of or saw a benchmark, the nviidia drivers were way ahead of nouveau in 3d video, games and such.

In addition the nvidia driver provides a nice configuration and admin interface that lets you easily change parameters like overall texture rendering quality, antialiasing, color correction, vsync etc.

But one of the best reasons to use the proprietary driver is that the settings utility lets you easily choose to change clocking on the fly.
Adaptive clocking vs full speed performance when gaming for instance. I havent tried the nouveau driver in a while, how does it handle adaptive clocking?

You can also monior temperature and memory use from the same utility.

All of the above are good reasons for using the nvidia driver imo, im not aware that the nouveau driver offers similiar tools and functionality, if so let me know and ill give it a shot

Found a recent benchmark:
https://www.phoronix.com/scan.php?pa...n-ubuntu&num=2
Seems the open source drivers are still lagging behind.

2 members found this post helpful.

Besides, nouveau driver doesn't support GPU computing, so it's pretty much useless for anyone intending to use its Slackware machine for some sort of scientific calculations or, more importantly, development of alike codes.

I don't know about the procedure of testing updates for Slackware packages, how many people are involved, and hardware availability to them, so I certainly cannot blame the team for not testing latest batch of updates with NVIDIA driver. However, with such slow Slackware release cycle, it's meaningless to claim that people should not complain about alike issues with -current. Because -stable is, frankly, not very interesting, and if not for -current, I guess lots of people would migrate away from Slackware anyway. And in such a situation, from an user point of view, it's really not that unreasonable to expect that, if possible, some important parts of infrastructure, like NVIDIA and maybe AMD proprietary graphics drivers, get tested with -current at least from time to time, even if these are not delivered along with Slackware.

3 members found this post helpful.