The 2010s Were Very Successful For Wine Thanks To CodeWeavers + Valve's Steam Play

[oiaohm]

Facts are what facts are, bottom line.... Wine is what it is and time itself has already proven what its doing doesn't work. EVERY single releases regresses badly... Wine unit tests are imaginary at best and wine devs refuse to emulate windows behavior and so long as that's true wine will continue to regress badly every single release.

Windows regresses every single release somewhere as well the wine test suite shows that as well. So regressing every release is emulating windows behaviour like it or not.

Also your claim that wine developers refuse to emulate windows behaviour is wrong and always has been. Funny enough the thing you are complaining about is windows behaviour.

[Cybmax]

Windows regresses every single release somewhere as well the wine test suite shows that as well. So regressing every release is emulating windows behaviour like it or not.

Also your claim that wine developers refuse to emulate windows behaviour is wrong and always has been. Funny enough the thing you are complaining about is windows behaviour.

Well, like it or not, i disagree with your statement that running "unit tests" = "real world apps". Simple. We do not see the winetests as the same.

I agree that running tests is a good thing, but to depend on those tests and claim "windows behavior", is clearly not "perfect". As you said in previous post, this is more of a "perceived behavior" kind of thing, and somewhat reverse engineered at that. No problem with that, its just seems at times that "this will not be implemented, cos it is not following X spec" or some shit is hindering stuff rather than promoting stuff.

Please, do not misunderstand me as being ungrateful cos i love wine.. its just that it seems very much a one-way-street (as with most developers). Sure, i understand that listening to a million ppl all spewing nonsense is not fruitful with limited time, but for the only way to ever get in touch with a dev is to use a mailing list? If you do not provide a thesis written explanation with long-tested code for some minor shit on the mailinglist, chances of it being reviewed is 0. Just look at all the code posted on the wine (and vkd3d) patchlist for wine currently. Some things are disregarded as "not important" if it is not linked to a particular bug, or if it is submitted by some "unknown" author. How many times do you think ppl will bother getting "their code in" if they just are met by silence and nothing more? Yeah, its a resources kind of thing... i know. It just makes stuff really hard, especially if some code is disregarded because it fails some winetest that ALSO fails on a real windows installation. Is it suddenly "right" to say that it is crap code, and counter the response that it ALSO does not work when run in windows with a claim that "your windows installation seems borked, because our reverse-engineered test is more correct than real windows behavior"?

Why do you think proton has made it so big so fast? Cos it's a "click here and run" thing, and not endless debates about how you installed it wrong, compiled it wrong, have the wrong hardware, use the wrong distro and so on and so forth. Its quite an hostile environment for "new" users to use wine, but that can also be said about a lot of other opensource programs. Using Linux is not for the faint of heart, cos the moment something does not work out of the box, you are met with ridicule, silence and plain rudeness... if you are lucky. Just look at the general tone of THIS board at times. Voice your opinion? Oh god.. lets tell him how stupid he is!

Yeah, that was a more ranted post than i wanted.. but i get frustrated at times. No point in debating this, cos we just do not agree. For me, a "real world app" is a "standalone running program". One program doing several functions != several apps for me.

[duby229]

Windows regresses every single release somewhere as well the wine test suite shows that as well. So regressing every release is emulating windows behaviour like it or not.

Also your claim that wine developers refuse to emulate windows behaviour is wrong and always has been. Funny enough the thing you are complaining about is windows behaviour.

No, the wine tests do -NOT- show that. All it shows is that neither wine nor windows -exhibits- wines tests for imaginary "correct" behavior.

Bottom line truth is most apps regress on wine every single release and as long as they keep on the path they are on it's gonna stay that way.

[oiaohm]

No, the wine tests do -NOT- show that. All it shows is that neither wine nor windows -exhibits- wines tests for imaginary "correct" behavior.

This is you just being wrong.
https://test.winehq.org/data/c84fa0a...dex_Win10.html
There is current a windows 10 test run that is 100 percent clean passed every single one of the tests in wine test suites So the correct behaviour is not imaginary. So windows sometimes does exhibit perfectly correct behaviour that all applications expect just it is not always and not as common and people would expect.

So yes from time to time windows installs 100 percent pass the wine test suite.

Bottom line truth is most apps regress on wine every single release and as long as they keep on the path they are on it's gonna stay that way.

Bottom line this is not true either. Please be aware I have access to the numbers so pushing the garbage is not winning you anything. I could give numbers on the regression rate its no where near most.

[Cybmax]

This is you just being wrong.
https://test.winehq.org/data/c84fa0a...dex_Win10.html
There is current a windows 10 test run that is 100 percent clean passed every single one of the tests in wine test suites So the correct behaviour is not imaginary. So windows sometimes does exhibit perfectly correct behaviour that all applications expect just it is not always and not as common and people would expect.

So yes from time to time windows installs 100 percent pass the wine test suite.

Now.. i am not about to claim a conspiracy theory here, but the submitter of the "0 error test" is [email protected]... And you have this little comment posted some time ago: https://www.winehq.org/news/2008121501

It's probably not that... but maaaaaaaaaaaaay look a bit suspicious

[oiaohm]

Now.. i am not about to claim a conspiracy theory here, but the submitter of the "0 error test" is [email protected]... And you have this little comment posted some time ago: https://www.winehq.org/news/2008121501

It's probably not that... but maaaaaaaaaaaaay look a bit suspicious

Yes the first 100 percent clean run on a windows machine was 2008. Please note I did mention that a 0 should be possible. Well before on turned up again. How come simple you are seeing machines with non aligned issues. So there is some combination where all of the tests will pass.

I was sticking to windows 10 because that what you had tested. Windows 7 two of it runs were 100 clean, 2008 has a clean run..

Windows 8/8,1 tests and the XP tests it looks close enough that if you did enough installs you would get clean run there as well. 2003 and Vista we don't have enough samples on.

A clean run of the wine test suite on real windows not exactly a one off these days it insanely common. Problem is we know that you remake virtual machine with the same version of windows and you can get slightly different results out the testsuite. Yes run on the single install instance wine test suite returns the same results over and over again. But you reinstall windows and you get different results.

So that one were game/program does not work and person reinstall their computer and it does work is explained by the wine test suite results.

https://test.winehq.org/data/287dabd9b6887e94cabfa2a5f9bfe822522095e5/index_Linux.html

When I say wine copies windows behaviour I really mean it notice that the wine test runs are not 100 percent consistent on wine. Also there is no way in hell in wine current state that is going to pass the wine test suite perfectly there is not even a chance by luck.

The difference between what wine test suite is checking in behaviours and what wine can in fact do is massive. Yes some of the breakages with wine between installs is replicating windows behaviour the bug for bug compatibility of the wine project..

[duby229]

Now.. i am not about to claim a conspiracy theory here, but the submitter of the "0 error test" is [email protected]... And you have this little comment posted some time ago: https://www.winehq.org/news/2008121501

It's probably not that... but maaaaaaaaaaaaay look a bit suspicious

Speaking of which, that's another major complaint I have about wine, it's developed specifically to expect the nVidia drivers non-standard opengl behavior. Run the wine tests on any mesa driver and you'll get dozens of failures even though the mesa drivers are nearly perfect implementations of the actual opengl specs and nVidia's drivers aren't even close. I'll admit that community members have improved wine for running on mesa drivers, but wine devs constantly break that work. That's really all that anomalous wine test shows is that they were specifically designed on and written for nVidia's proprietary drivers non-standard opengl implementation.

It's just another reason why I think that what those wine tests actually test for is entirely imaginary.

[Cybmax]

Yes the first 100 percent clean run on a windows machine was 2008. Please note I did mention that a 0 should be possible. Well before on turned up again. How come simple you are seeing machines with non aligned issues. So there is some combination where all of the tests will pass.

The post from 2008 i pointed to kinda indicate that it is only 1 (developer run) machine that actually are able to (puzzling and poorly understood phenomena) run the winetest "perfect".

This (in case you actually did not bother to try to understand) COULD indicate that the only machine able to actually run these tests are 1 "perfectly set up to work with the winetest suite" windows install (something the report coming from [email protected] would indicate). So what does this mean? Yeah, the tests that tests for "perfect windows functions" ONLY works if your windows is installed JUST like that one, with the SAME hardware/drivers/version of EVERYTHING.

Now, please inform me if that is a good way to test something?

[oiaohm]

Speaking of which, that's another major complaint I have about wine, it's developed specifically to expect the nVidia drivers non-standard opengl behavior.

This is really not as simple as you are make out. Turns out Nvidia implemented some stuff in opengl first that is required to in fact emulate dx. Also these Nvidia first stuff is also required so windows opengl games in fact work.

Run the wine tests on any mesa driver and you'll get dozens of failures even though the mesa drivers are nearly perfect implementations of the actual opengl specs and nVidia's drivers aren't even close.

This is speaking out your ass today was historically correct. The best test runs of the dx areas of wine at the moment and moment being 4 years are all on mesa drivers. Nvidia is more broken than Mesa to the wine test suite these days. So Nvidia uses should be quite upset at the moment and mesa uses should be quite happy.

I'll admit that community members have improved wine for running on mesa drivers, but wine devs constantly break that work.

This is also false for the last 5+ years. Why you submit patch to wine it will be automated tested by the winetest bot that uses mesa heavily since a particular change in the testbot and if you fail by testbot you patch is rejected until you get it to pass again. So wine developers cannot constantly break mesa driver support for the past 5+ years as the wine development process does not allow it. This is something historic that is no longer true and has not been true for quite some time..

That's really all that anomalous wine test shows is that they were specifically designed on and written for nVidia's proprietary drivers non-standard opengl implementation.

What you have written is inverse of current reality. The anomalous wine tests on Linux systems with Nvidia currently shows wine is written more for mesa drivers than nvidia ones.

Yes wine test suite is run on windows machines with Nvidia cards yes this does cause higher failure rates there that I will explain in the Cybmax section of this answer. Please do remember a lot of windows games are only built and tested with Nvidia as well so our VM of windows has to run Nvidia to see exactly what behaviours applications are expecting.

The post from 2008 i pointed to kinda indicate that it is only 1 (developer run) machine that actually are able to (puzzling and poorly understood phenomena) run the winetest "perfect".

The phenomena is understood but understanding really does not help you. You have to remember wine test suite is based on the behaviour applications expect to see to run correctly. Of course until wine started doing test suite no one thought for one min that windows would be doing stuff to API/ABI that broke application compatibility.

ONLY works if your windows is installed JUST like that one, with the SAME hardware/drivers/version of EVERYTHING.

This is in fact wrong and this is something the wine test suite has shown. "SAME hardware/drivers/version of EVERYTHING" this is not enough to make sure that you get the same result out of winetest. Yes this was mega hair pulling that you can have that all the same have 2 installs of windows and get 2 different results out wine test suite.

At first when the wine project was not running multi instances of windows to check stuff caused developers to think tests were just wrong and lead to a few cases of one developer fixed a test one way then another undo it because it believed to be a regression when the real problem was a real windows problem.

This is one area where duby299 is kind of right. Some of this hair pulling is drivers like Nvidia when you install the driver it probes the hardware and alters it settings based on what that probing return then settings in the registry to be used from then on. So each time you install drivers like this it is like running a random number generator on how those settings are set. Only way to rerun that is fully uninstall the drivers and reinstall the driver(the driver makers uninstall tool will not remove those registry settings) or do a full clean install.

its not just the drivers playing a random game with the ABI stability windows activation(yes Microsoft version of digital rights management) is really fun that you can have two different wine test suite results before and after activation.

Now, please inform me if that is a good way to test something?

You are running test to confirm that your theories are right or wrong. Sometimes horrible answers are what you get.

The answer wine project has got is windows ABI with everything is about 90 stable per Windows version. That leaves 10% of random hell worse this 10% hell is not aligned between Windows versions.

This hell makes people think looking at the wine testbot results that wine tests have to be wrong when they are not. The problem is Windows is horribly wrong and wine need to-do the behaviours applications expect.

[duby229]

This is really not as simple as you are make out. Turns out Nvidia implemented some stuff in opengl first that is required to in fact emulate dx. Also these Nvidia first stuff is also required so windows opengl games in fact work.



This is speaking out your ass today was historically correct. The best test runs of the dx areas of wine at the moment and moment being 4 years are all on mesa drivers. Nvidia is more broken than Mesa to the wine test suite these days. So Nvidia uses should be quite upset at the moment and mesa uses should be quite happy.



This is also false for the last 5+ years. Why you submit patch to wine it will be automated tested by the winetest bot that uses mesa heavily since a particular change in the testbot and if you fail by testbot you patch is rejected until you get it to pass again. So wine developers cannot constantly break mesa driver support for the past 5+ years as the wine development process does not allow it. This is something historic that is no longer true and has not been true for quite some time..


What you have written is inverse of current reality. The anomalous wine tests on Linux systems with Nvidia currently shows wine is written more for mesa drivers than nvidia ones.

Yes wine test suite is run on windows machines with Nvidia cards yes this does cause higher failure rates there that I will explain in the Cybmax section of this answer. Please do remember a lot of windows games are only built and tested with Nvidia as well so our VM of windows has to run Nvidia to see exactly what behaviours applications are expecting.



The phenomena is understood but understanding really does not help you. You have to remember wine test suite is based on the behaviour applications expect to see to run correctly. Of course until wine started doing test suite no one thought for one min that windows would be doing stuff to API/ABI that broke application compatibility.


This is in fact wrong and this is something the wine test suite has shown. "SAME hardware/drivers/version of EVERYTHING" this is not enough to make sure that you get the same result out of winetest. Yes this was mega hair pulling that you can have that all the same have 2 installs of windows and get 2 different results out wine test suite.

At first when the wine project was not running multi instances of windows to check stuff caused developers to think tests were just wrong and lead to a few cases of one developer fixed a test one way then another undo it because it believed to be a regression when the real problem was a real windows problem.

This is one area where duby299 is kind of right. Some of this hair pulling is drivers like Nvidia when you install the driver it probes the hardware and alters it settings based on what that probing return then settings in the registry to be used from then on. So each time you install drivers like this it is like running a random number generator on how those settings are set. Only way to rerun that is fully uninstall the drivers and reinstall the driver(the driver makers uninstall tool will not remove those registry settings) or do a full clean install.

its not just the drivers playing a random game with the ABI stability windows activation(yes Microsoft version of digital rights management) is really fun that you can have two different wine test suite results before and after activation.



You are running test to confirm that your theories are right or wrong. Sometimes horrible answers are what you get.

The answer wine project has got is windows ABI with everything is about 90 stable per Windows version. That leaves 10% of random hell worse this 10% hell is not aligned between Windows versions.

This hell makes people think looking at the wine testbot results that wine tests have to be wrong when they are not. The problem is Windows is horribly wrong and wine need to-do the behaviours applications expect.

Another wall of bullshit, but what you just did in that wall of bulshit was make another good argument for how the wine tests are nonsense. First of all, every single time anybody has ever come anywhere close to 100% passing every test was done on nvidia hardware. The problem with that is that nvidia hardware is not compliant with opengl specs. Second of all due to wines "one behavior to rule them all" philosophy that specifically means that anything outside of expectations, which includes everything that's not nvidia, basically everything that is standards compliant, will fail tests.

Third of all your point that applications will often times run different code paths depending on what hardware is detected also makes a great argument for exactly why wines tests are synthetic and don't make any real world sense.