The Significant Corporate Importance & Pressure Around Mesa Open-Source Linux 3D Drivers

[schmidtbag]

zlib isn't both backwards and forwards compatible as this would literally mean not adding any new functions. Given this, software using zlib should not use any new functions that are added after a minimum supported version and thats precisely what mesa did. Mesa cannot control what version of zlib is loaded.

Unless the new functions are fundamentally incompatible with the old ones, I don't see why they can't both exist at the same time. Basically you'd just have legacy code that doesn't change while anything expecting the new version will use the modern code.

A physical analog is diesel engines. The original diesel cars can run on both modern fossil-based diesel fuel at the pump as well as straight vegetable oil. Meanwhile, modern diesel vehicles can only operate on the modern stuff. In this analogy, the old tech has wider compatibility without really hindering the new stuff.
So - seeing as there's two distinct functions in zlib (crc32 and crc32_z), why can't both exist in the code and are just called upon as needed? Mesa could still have zlib 1.2.9 as a minimum dependency but in keeping crc32, SPEC can still run.

Of course, having not worked with the code for zlib (or Mesa), there may be a reason why both crc32 and crc32_z can't exist, I just find it a little odd that this is the case since they have distinct names.

[sophisticles]

Windoze fans are far more cult like than the Mac. And always have been. They just don't recognise it with heads buried in the sand is all.

This is rich coming from a Linux advocate.

Linux use encompasses a number of different groups, for instance people like me that use both Linux and Windows, see a need for both but also see weaknesses in each and strengths in each.

Then you have the subset of Linux users that demand everything be GPL, for who this is a religion, just like any other formal religion, with a blind adherence to ideology.

These people are way more cult like than anyone else and they tend to simultaneously be the least knowledgeable.

[sophisticles]

We may think whatever we want about open source ideology and software development, but this proves that Linux now has actual business on the desktop

If the Linux fanatics that demand everything must be GPL weren't such obstructionists, Linux would have a much bigger presence on the desktop than it has now.

[sophisticles]

They've always had a policy of making macs that'll never boot any OS version older than what was current when they were finalized... e

In all fairness, it's ridiculous to expect any company to support an OS version that is older than what that specific model is designed to run with.

Imagine Dell selling a new laptop that was designed with Win 11 in ming and being forced to support installing Vista or XP on it.

I can understand from the standpoint of wanting to head off all the support calls such a thing would result in for them to make it so installing an OS older than what was shipped is impossible,

[sophisticles]

Not really. Ability to learn from the open source code was certainly a side affect but in reality it's mainly these two reasons:

a) There was a high diversity in hardware (and OSes for that matter) without some sort of established de-facto "industry standard", so open source was really the only form of software which was practical to ship;
b) There was no concept of a "software as product" (and "software company") back then, so naturally binary form of a software had no sense;

https://en.wikipedia.org/wiki/History_of_free_and_open-source_software

[sophisticles]

I think a good litmus test which decision is good for your FOSS project is WWLD ("What would Linus do?").

We know what he would do, he would talk a good game but take 1.7 million dollars a year while doing so.

Linus is just like any of the sham preachers you see on TV, preaching Christianity while reaping on millions from suckers, i mean believers, and living in the lap of luxury.

On the other hand, any decision which impedes progress and only serves business interests, I think Linus would have some choice words for that.

Those words would ring truer if he wasn't all too happy to rake in millions from corporations that contribute in furtherance of their business practices.

Linus is a true communist, a do as i say not as I do type of guy.

Make no mistake, i envy him, he has convinced millions of people that he is a saint all the while laughing to the bank.

Good for him.

[sophisticles]

You got confused by how the word "free" is being used in different contexts.

Don't worry, many other people who were, like you, new to the concept of FOSS fell into that misunderstanding.

The word "free" in this context means that the recipient of the software is free (to a certain degree) to do with the software what they want.
it does neither mean that the software came without cost or that the developers creating it were not paid to do so.

Given that you are posting on a site about Linux topics you are probably aware of the Linux kernel.
It is FOSS and its principle engineer Linus Torvalds is paid quite handsomely to develop it.

Another good way of getting one's head around this is to consider an in-house software which is later released as FOSS.
In this case the employees who created it were paid their salaries and continue to be paid the same salaries after the release.

The release has simply expanded the people who get access to the software (first company internal, then beyond that) and releases as FOSS expanded what these people can do with the software once they have it.

There have been many attempts at replacing the word with something less ambiguous but few of them have gained traction.
it is also often less ambiguous in other languages.

I can tell you do not know what my background is and I am not going to list my resume here because i don't want people think that I am bragging or as one person claimed that i was a narcissist.

I will tell you that I am a certified Unix Sys Admin from a major university and I have 2 college degrees on top of that.

I have been running a Windows/Linux dual boot for decades, going back to Win 2k/Red Hat on a 900mhz Athlon.

I have heard variations of what you have stated for decades and they all fail to explain how the revenue is generated is the software is given away for free.

Linus and the Linux Foundation are an outlier because they get their money from the corporate sponsors, like Intel and Microsoft, who makes billions selling proprietary products.

But assume you just graduated college, have a degree in comp sci, possibly hundreds of thousands of dollars in student loan debt and you decide to work on a project where the source code is being given away for free.

Where exactly do you think the money is going to come from?

The KDE project is not rolling in dough.

The XFCE project is not rolling in dough.

In fact, most projects are not able to generate any revenue other than from donations, or a nicer way of saying it, sponsorships.

Show any economic theory, any business theory, or even basic math principle, where you can give a product away for free and still generate revenue.

Look at websites, how do they generate revenue? Subscriptions or ads or both, without that they don't make money by giving away their content for free.

If a book author gives his book away for free how does he make money from it?

I'm going to assume you don't work for free and I assume your employers isn't giving away their services for free, so why should programmers work for free?

[hsci]

Finally some sense.

But in the long run, software that bundle their own libz.so should note this and provide an FAQ or KB informing people about this, and eventually update their code to work with a newer Mesa and libz.so.

And if it's not possible, the proper way for Mesa should be to do a

if
{
dependency('zlib', version : '> 1.2.3 && <= 1.2.5')
foo
}
else if
{
dependency('zlib', version : '>1.2.5')
bar
}​

Brutal truth: a lot of what the so-called FOSS developers call shitty code quality in proprietary code is actually an extremely comprehensive set of layers upon layers of dependency versioning catches to ensure the program works acceptably against a shitton of older and newer dependencies that may exist in the variety of user systems and configurations. But this is something the so-called FOSS developers bragging about their so-called code quality will never ever be able to understand.

Generally: Yes. And Mesa compatibility and performance matter for all of us. Whether developer (paying bills) or user (using software).

This kind of code is also hard to maintain (old GNU code with thousands of special paths for that UNIX, BSD, Nextstep...). Backwards compatibility is required, as soon as it becomes bug compatibility and workarounds for other software (elsewhere) it becomes a burden. And doing the Apple thing (e.g. just dropping compatibility) isn't a thing. Linux and a lot GNU software has driven compatibility code to an art level - named quirks. So I wouldn't be surprised about:

vector<quirk> quirks = load_quirks(executable, environment);
if (!compatibility_check_quirks(quirks)
    exit(1);
exec_quirks(quirks);

What I miss in the article:
It describes the cons of an zlib upgrade, not the benefits.

PS: Valve is doing compatibility on a high level with "Proton" but that isn't actually their desired work. They want native executables and care about HL, CS and Steam. At some point Proton shall become superfluous or Valve will doing the job of others.

[Old Grouch]

If the Linux fanatics that demand everything must be GPL weren't such obstructionists, Linux would have a much bigger presence on the desktop than it has now.

And 'Linux' would be almost, but not quite, entirely unlike the Linux we know today. Nutri-Matic Linux.

[ezst036]

This is conspiracy style thinking and you should look at yourself as such.

While its true that shady business practices are a factor, its only one of many and it also doesn't explain why Linux dominates in other areas (i.e. servers) where companies using Linux have also arguably used shady business practices (google/redhat anyone). Business is business, regardless of whether we are dealing Linux or other OS's.

It's not really conspiracy thinking at all. Windows got on top in the era of 95/98 and had already been established(partially due to DOS lineage), and was dominant during the XP era. Just based on market numbers that sort of establishment isn't done away with overnight. The shady business practice that comes into play is Microsoft's well known subsidy to vendors(HP, Lenovo, etc) to keep Windows on their machines. Who's going to switch to Linux to save $5 dollars? <-- I'm not kidding with this question.

The issue is "default", and Microsoft learned a long time ago (and got sued over it[Internet Explorer]) the raw power that comes with being the default. It's not a conspiracy at all. It's what is. When your'e the default, you are "what is", you are "the fact". That's not a conspiracy.

it also doesn't explain why Linux dominates in other areas (i.e. servers) where companies using Linux have also arguably used shady business practices (google/redhat anyone). Business is business, regardless of whether we are dealing Linux or other OS's.

If business licenses for Windows servers were $50 dollars each or less Linux never would've taken off. Likewise if home users had to pay thousands of dollars per home Windows license, the reverse would be true. Linux would dominate the desktop and would be crap on servers that nobody would use because the licenses were dirt cheap.

Note that Linux decimated Unix before it took out Windows servers because while Windows server licenses are hundreds/thousands of dollars, Unix licenses used to be thousands/tens of thousands of dollars - plus the ease associated with Unix users switching to Linux. The ease of switching Windows servers to Linux does carry with it a tech knowledge hurdle to overcome, but that isn't enough in the face of licensing fees. Who's going to switch to Linux to save $10,000 dollars?(or more) - plus the low cost of off-the-shelf hardware vs the massive cost of proprietary mainframes. Now you have a good question with a good answer. That's why Linux took over servers. Follow the money.

There still are many lower-end windows servers out there most notably associated with active directory. The cost really isn't exorbitant so there's less incentive. Linux will kill it too, eventually. The bad economy will ensure that.

The core issue here, which other people have mentioned in this thread is that historically Linux has placed less import on not breaking users (usually this is more of a userspace issue then the Linux kernel issue, Linus is famously particular about this) but for users the distinction is not relevant here.

Ontop of this you also have fundamental decision decisions in Linux which other OS's such as Windows/Mac don't have which objectively makes it harder for users especially for desktop which is the **one** area where Linux has pathetic traction, i.e. Linux being a monolithic kernel which means that it doesn't have stable ABI's for things like graphics card drivers which matters when you know, you buy a new graphics card and you need the latest driver but for obvious reasons you don't want to be forced to use the latest version of Linux kernel (which tend to less tested than older more stable versions).

This problem is less of an issue in closed down systems where the vendor has almost complete control over the hardware (i.e. steam deck, android phones) but you can't avoid this problem with laptop's/desktops where the hardware configuration can be anything.

On the whole I agree with all of this. The different model has to be treated differently. Some of this isn't deal breakers though. On distros which have a huge amount of users, bug testers catch this stuff. I've used specialty distros as well as the most used like Ubuntu and Mint. Stability and a huge user base very frequently come hand in hand.

Many eyes.

Your point about graphic card drivers is very well stated.