Steam stops supporting Windows 7 and Windows 8

[LexiMax]

12 minutes ago, segfault said:

The real problem with depreciating operating systems is that it's forcing people into new operating systems that are objectively fucking terrible.

 

And that's the most frustrating part.

 

Linux could have reasonably been that alternative if the Desktop Linux community could have stopped being its own worst enemy for two seconds.  It has had opportunities, but the problems preventing widespread usage on the desktop are footguns of their own creation, and not evil plots by Microsoft, Apple or Google.

Edited May 4, 2023 by LexiMax

[Murdoch]

1 hour ago, Mr. Freeze said:

Clearly the answer to a slight inconvenience provided by Windows is to suffer through the pain of switching to Linux 

 

Linux has it's uses, but it is not prime time ready Joe Average User. People with technical experience often forget what it's like for inexperienced users. As a computer tech, I deal with that all the time. I tried to install Linux for people twice over the last couple of years. The first was an Asus laptop where it had a problem of it not booting properly without being plugged into power. That took some research and a kernel update to fix. The second was a desktop build where multiple distros failed to install on an M.2 card no matter what I did, even after much research and screwing around with BIOS settings. Windows installed on it fine, so there was nothing wrong with the card or board. Ended up reverting to a SATA SSD in a desperate attempt to salvage the job which, thankfully, worked. Windows has many, many issues, but the simple fact is it more often than not just works for the overwhelming majority of use cases. And that works best for most people including myself.

[Andromeda]

2 hours ago, Individualised said:

Like I actually can't believe they're shipping an operating system that performs this poorly on modern hardware. You wait about 5 minutes to perform simple tasks, it's unbearable to use.

2003 isn't modern anymore.

[Individualised]

2 minutes ago, Andromeda said:

2003 isn't modern anymore.

What? I'm not sure where you gathered I was using 2003 hardware from, but I'm not? Very strange post!

[Blzut3]

1 hour ago, LexiMax said:

It has had opportunities, but the problems preventing widespread usage on the desktop are footguns of their own creation, and not evil plots by Microsoft, Apple or Google.

As someone who distributes proprietary Linux software for my day job, I may disagree strongly on the specific degree to which it is actually a problem.  (To be fair it can vary a bit depending on the specific type of software.)  But I do agree that with some fairly small policy and attitude changes it could be a lot better than it is today.  "Footguns of their own creation" is a very succinct way of putting it.  Many of them only exist because of absolutism/perfectionism.

[Murdoch]

4 hours ago, Individualised said:

What? I'm not sure where you gathered I was using 2003 hardware from, but I'm not? Very strange post!

 

Because what you are describing is not even close to normal. All of my Windows 10 machines boot to usability in around a minute. Hell even the basic-ass dual core CPU 8GB RAM laptops I sell my "just the basics please" customers load Windows 11 in probably 90 seconds give or take. So I do not know what is going on with the Windows 10 machines you have used, but five minutes to usability is not even close to standard. I have added SSDs to 7 - 8 year old desktops and got better boot times than that with Windows 10.

 

The only time I have seen behaviour like you describe is Windows 10 trying to load on a deteriorated mechanical hard drive. They are like engines. They get worn out and inefficient as they age. Windows 10 is definitely more demanding of drive performance than 7, so any degradation related performance problems are going to be made that much worse in an upgrade from 7 to 10. I would go so far as to say Windows 10 (and presumably 11 by extension but I haven't tried it) has absolutely no business running on a mechanical hard drive. A couple of years ago two different customers of mine made the mistake of buying the same Windows 10 machine with mechanical hard drives without consulting me first, and the performance was absolutely atrocious even on the presumably brand new drives.

Edited May 5, 2023 by Murdoch

[Edward850]

13 minutes ago, Murdoch said:

Windows 10 is definitely more demanding of drive performance than 7

I suspect it isn't that it's more demanding, but rather they've changed optimization targets. SSDs are the new norm, even if just used as the OS drive, so they are targeting operation around those now.

[DrinkyBird]

Of all the complaints I have about Windows 10, performance isn't really one of them. Even on a low-end cheap SSD it ran fine. Hell, I used it fine on a not-so-old-at-the-time spinning disk hard drive. So if it runs terrible then either your hardware is archaic or you installed it on slow-ass storage.

 

And Linux isn't invulnerable to slow storage either... I've had the entire OS hang up while doing I/O-heavy work when running Linux off a hard drive.

[heliumlamb]

14 hours ago, Gez said:

How so?

 

KHTML itself is long dead, having had its last updates in 2016. Apple's WebKit was forked from KHTML in 2001, and Google's Blink was forked from WebKit in 2013. Safe to say that they've been doing their own thing for a looonnnng while now, and that they would have also done their own thing if they started from Gecko.

you want to know how so?

what is a fork if not a consequence? foul roots are still roots no matter the appearance of that which has grown from them

 

[nolongeramnion]

 

Spoiler

I hate bing so much

 

[Edward850]

3 minutes ago, amnion said:

 

Counter Strike 2 has no listed system requirements. You are looking at the system requirements someone copied from CSGO for an SEO-bait article, and CSGO was released in 2012.

[Professor Hastig]

7 hours ago, Blzut3 said:

As someone who distributes proprietary Linux software for my day job, I may disagree strongly on the specific degree to which it is actually a problem.  (To be fair it can vary a bit depending on the specific type of software.)  But I do agree that with some fairly small policy and attitude changes it could be a lot better than it is today.  "Footguns of their own creation" is a very succinct way of putting it.  Many of them only exist because of absolutism/perfectionism.

 

My take here is that if someone really wants to they will find a way. At the end of the day we still have to cope with a setup that ironically has a lot in common with how big tech corporations try to gatekeep the access to their systems: On iOS we got the App Store, on Android we got Google Play and on Linux we got the distro's repository. If something isn't in there or only as an outdated version it can become quite a hassle to get things working. It's by no means impossible to get around this but surely not straightforward anymore.

 

You can surely claim that Windows's installers are not the best way to distribute software - but they are very efficient at allowing free distribution through any desired channel without making the user suffer some inconveniences.

 

Macs once had that, too, but the increasing lockdown of the system has certainly resulted in narrowing the usable distribution channels.

 

 

The gist of it all: If you want to compete with Windows, it is essential to compete with its software distribution options. And Linux is absolutely not there right now. If it was it'd be a lot easier for commercial entities to sell their software for the platform which is an absolute requirement to gain some market share.

 

[Gez]

7 hours ago, heliumlamb said:

you want to know how so?

what is a fork if not a consequence? foul roots are still roots no matter the appearance of that which has grown from them

I get it: you blame Apple's and Google's behavior entirely on KDE having existed.

[heliumlamb]

2 hours ago, Gez said:

I get it: you blame Apple's and Google's behavior entirely on KDE having existed.

Of Course, I Exactly Said That It's KDE's Fault That Google's Behavior Is What It Is From My Computer Running Plasma 5.27 Of My Own Volition

Edited May 5, 2023 by heliumlamb

[Gez]

2 hours ago, heliumlamb said:

Of Course, I Exactly Said That It's KDE's Fault That Google's Behavior Is What It Is From My Computer Running Plasma 5.27 Of My Own Volition

Indeed, since you blamed the "foul roots" which were KHTML...

[heliumlamb]

2 minutes ago, Gez said:

Indeed, since you blamed the "foul roots" which were KHTML...

We Live In A Society

[BBQgiraffe]

20 hours ago, Individualised said:

@BBQgiraffe was implying that there was something wrong/Windows-like with recent versions of Ubuntu, that is what I was asking about.

yup, most of the software I was using on Windows still works fine or has a drop-in replacement

[BBQgiraffe]

16 hours ago, Graf Zahl said:

Or: Wanna use version 2.25 of some library? You betcha that some distro sticks to 2.20 for whatever dumb reason.

In such an environment you become dependent on people you can't control and can't convince that their way is not ok.

The most ridiculous thing I experienced was that one Linux distro made changes to the GZDoom project to exclude a few libraries we included in source code because we had to make some changes, just to revert to the system provided variants that do not work correctly. Needless to say, I refuse to give any tech support to such versions. The noise of these things still sticks around, though.

 

 

god yes, I had to build Xash3D on an old Lubuntu version due to 32-bit being dropped a while back, and ended up needing to compile like 10 or so massive libraries because one of the main dependencies was like version 0.13.7 and apt only let me download 0.13.6 or something, and then that repeated for all the dependencies of that dependency. I don't have a lot of complaints about Linux, but distro maintainers sticking to old package versions or not updating their 32-bit packages(especially Lubuntu, considering their main thing is that they run on old machines) is probably #1 on that list

[LexiMax]

8 hours ago, Professor Hastig said:

on Linux we got the distro's repository

 

And it's somehow even worse, because at the very least when I distribute my software in an app store the expectation is to vendor my dependencies.  Linux distros want my software to become so intertwined with their OS that all of my dependencies are swapped out for their versions, so if someone files a bug I can't rule out mismatched dependency version as a cause.

 

That's insanity.  I cannot emphasize enough that it's literally the worst of both worlds and the dumbest way to distribute software possible.  If I release a program that depends on SDL, I should not have to prevent myself from using - say SDL_RenderGeometry - just because some Linux distro is stuck on an older version of SDL before 2.0.18 that doesn't have that function.  Or the opposite - maybe I'm using an old version of a library because the newer versions past might introduce spurious compiler errors or differing functionality that I don't have to worry about until I decide it's worth upgrading.

 

You want to distribute my software for me and with your own changes to integrate it into your OS?  You support it.  But if that's the only popular way to distribute software on Linux, that's a problem.

Edited May 5, 2023 by LexiMax

[HavoX]

19 hours ago, ENEMY!!! said:

I think that's a major reason why Microsoft was keen to push compulsory automatic updates with Windows 10.  Their official line was that it was for security, but another major factor was that with Windows 7/8, they couldn't get away with using automatic updates to do things like reinstalling Edge when it is removed, because it could lead to more users turning off automatic updates and not regularly updating manually, which would present security problems.

Fair enough.. BUT STILL.

 

As for automatic updates, I just follow AskWoody's directions and defer updates when told to.

Edited May 5, 2023 by HavoX : Might not want to get political...

[Graf Zahl]

35 minutes ago, LexiMax said:

 

And it's somehow even worse, because at the very least when I distribute my software in an app store the expectation is to vendor my dependencies.  Linux distros want my software to become so intertwined with their OS that all of my dependencies are swapped out for their versions, so if someone files a bug I can't rule out mismatched dependency version as a cause.

  

That's insanity.  I cannot emphasize enough that it's literally the worst of both worlds and the dumbest way to distribute software possible.  If I release a program that depends on SDL, I should not have to prevent myself from using - say SDL_RenderGeometry - just because some Linux distro is stuck on an older version of SDL before 2.0.18 that doesn't have that function.  Or the opposite - maybe I'm using an old version of a library because the newer versions past might introduce spurious compiler errors or differing functionality that I don't have to worry about until I decide it's worth upgrading. 

  

You want to distribute my software for me and with your own changes to integrate it into your OS?  You support it.  But if that's the only popular way to distribute software on Linux, that's a problem.

 

That's essentially the core of the problem. I simply cannot offer support for such a way of distributing software. Especially as a commercial developer I only can do that as long as I can be certain that the correct library versions get used. Both Windows and macOS offer that guarantee by default - but I cannot give any guarantees whatsoever if I am at the mercy of a repo where things can change on a whim and in really bad cases dependencies can be pulled, because some library I need gets replaced with a new API/ABI breaking revision.

In a commercial environment that's unfeasible. Linux's way to 'address' this is LTS releases but all they do is not update the libraries (and compilers!) at all, so I am screwed squared now.

 

 

[Blzut3]

32 minutes ago, LexiMax said:

I should not have to prevent myself from using - say SDL_RenderGeometry - just because some Linux distro is stuck on an older version of SDL before 2.0.18 that doesn't have that function.

So because Windows 7 Win32 API doesn't have functions that Windows 8+ does you shouldn't support Windows?  We saw with the 9x to NT and XP to Vista transitions what happens when you assume you can just dynamically link to Win32.

 

SDL has been basically the gold standard when it comes to ABI compatibility.  Besides a couple mis-steps like 2.0.6 (but of course nothing is perfect), it's generally pick the minimum you wish to support and whatever version the user is using should work fine (at least that's been my experience).  Recently they even stepped up to supporting the previous ABIs with wrappers to the new version.  Early in the 2.0 cycle they even added the ability for the user to override a statically linked copy if they desire, so you can totally statically link SDL to ensure a baseline and if some new sound system or windowing system comes out the user can still replace SDL to try getting native compatibility.

 

While I think the whole "they can patch the library under me" thing is overblown, I will admit that it's often hard to discern what the ABI policy is for a given library.  Those who hold an absolutist view will encourage dynamic linking to stuff like libjpeg which changes ABI frequently, or worse stuff which has no ABI compatibility guarantee whatsoever (as Graf complains about), and yes all of this being inconsistent and per project makes it hard to do Linux binaries well.  Then many distros will only provide the latest ABI and discard any backwards compatibility with no transition period because of security concerns making things worse.

 

A better example of how the traditional distro model needlessly makes things painful is to look at how many of them do not offer GPU driver updates (specifically the user space side of things), and for many of them there's not really an endorsed way to do that.  While there are usually ways to do it that are not necessarily hard, they do have poor enough user experience to discourage it by design.  Static linking and containers can't really help here either since we'd be back to the DOS days where software had to support the hardware.

 

Ultimately I don't want to get too deep into a debate here since this is one of those areas where people don't really change their opinion.  All I can say is that there is often a double standard when it comes to distributing Linux software, probably because Linux users can be very vocal.  On Windows and macOS you'll say version X is less than 2% of the user base for that OS so we'll drop it.  On Linux the discussion always comes to "what if someone has this ultra rare arcane setup, then what?"  Again, I do admit there are lots of problems here, but I think there's a lot more nuance to which things actually make life difficult vs which are just "I'm not familiar with this and 2% of the total addressable market isn't worth figuring it out."  Granted as said earlier, one wouldn't need to figure out many of these things if the distros had slightly different policies.

[Graf Zahl]

8 minutes ago, Blzut3 said:

Granted as said earlier, one wouldn't need to figure out many of these things if the distros had slightly different policies. 

 

I can't agree that 'slightly different' policies would help. Yes, you can somehow cope with the library mess - but the entire setup is not what commercial software needs. AppImage and Flatpak are more like it, but they still seem to be met with too much resistance.

 

[Wadmodder Shalton]

Next, when Steam ends support for Windows 10 in a few years after that operating system's end of life in 2025, it would be the end of Steam on Windows 32-bit systems as far as we know.

[dpJudas]

1 hour ago, Blzut3 said:

While I think the whole "they can patch the library under me" thing is overblown, I will admit that it's often hard to discern what the ABI policy is for a given library.

For me it isn't so much if they patch the library under me, but rather that I've come across several cases of where they've compiled the library incorrectly or excluded features from the library that I require. For example, a large distro had a version of LLVM where the CMake files were basically broken. In a different case freetype2 was built with things excluded that I needed. Even if I contacted the distro maintainer and managed to convince them to fix it (which in itself might not be successful), I have to wait years afterwards before the complaints end in the support channel.

 

Why does my users have to build my program? Well because otherwise they have to wait years for the next complete distro update to get next version. Assuming the next distro version even syncs to a new version of it.

 

Then there's the case with asmjit where I had to patch the library for call stacks to work. Here the asmjit author wanted me to do additional work before merging it as a PR. If the distro insists there can only be the asmjit author's version, then that effectively means call stacks for ZScript JIT errors for the users are thrown under the bus. Sucks to be them, right?

 

UDB's mono version suffers from issues with the mono implementation of winforms. If I actually went and fixed some of those the distribution model means users would have to wait years before they ever get them. Distros don't seem to pull mono changes from upstream particular often.

 

This isn't a serious way to develop software IMO.

[Edward850]

6 hours ago, Wadmodder Shalton said:

Next, when Steam ends support for Windows 10 in a few years after that operating system's end of life in 2025, it would be the end of Steam on Windows 32-bit systems as far as we know.

So little people use Windows 10 32bit right now that Steam doesn't even have a record for it in their hardware survey. 32bit makes up less than 0.1% of hardware right now. People have already moved on from it.

[Blzut3]

6 hours ago, Wadmodder Shalton said:

end of life in 2025

Just being a pedant: Windows 10 won't fully go EOL until at least 2032 (actual final date is still TBD) for customers that really want to pay for it at least.  Of course those same customers would tell you Windows 7 is still supported for another year and half-ish.  (Fun fact: Actually had someone inquire at work if we supported the embedded version of Windows 7 today.)

 

Will definitely be interesting to see how fast Windows 10 gets dropped from browsers (and thus Steam) when 2025 rolls around.  Although I suspect it might be awhile longer given the whole situation with Microsoft artificially locking Windows 11 to newer hardware.

 

As Edward850 said though 32-bit isn't a concern since basically no one uses 32-bit Windows 10, and many hardware vendors don't even build drivers for it.

 

[Murdoch]

6 hours ago, Wadmodder Shalton said:

Next, when Steam ends support for Windows 10 in a few years after that operating system's end of life in 2025, it would be the end of Steam on Windows 32-bit systems as far as we know.

 

What is your obsession with 32 bit? I really don't get it. It's been dying off for probably 15+ years, and except for vintage enthusiasts is no longer a going concern. And they'll find a way to keep what they want running. They always do. No one is releasing 32 bit only hardware in the mainstream market.

[Murdoch]

7 minutes ago, Blzut3 said:

Will definitely be interesting to see how fast Windows 10 gets dropped from browsers (and thus Steam) when 2025 rolls around.  Although I suspect it might be awhile longer given the whole situation with Microsoft artificially locking Windows 11 to newer hardware.

 

I won't be surprised if they are forced to extend support past 2025. Plenty of still viable gear can't run 11 and ms is living in lala land if they think people will just go out and buy new computers en masse.

[LexiMax]

7 hours ago, Blzut3 said:

So because Windows 7 Win32 API doesn't have functions that Windows 8+ does you shouldn't support Windows? 

 

Windows doesn't ship SDL.  If you want to use SDL, you vendor it.  Windows is set up for this to be the easy and expected way you handle your business on the platform.  This approach does have consequences - for example, I have 87 distinct copies of SDL2.dll on my system.  But do I care?  Not really - my programs still work.

 

Perhaps Linux distros shouldn't ship the entire open source ecosystem in its package manager if they can't fight their urge to tie dependencies together in a neat little bow.

Edited May 6, 2023 by LexiMax