I've always chosen not to install the "preview" updates, and always been a little surprised that Microsoft offers them at all. Isn't that what beta testing is for?
> Didn’t they fire all their in-house QA ages ago?
No, they didn’t? Like most other companies 10 (or so) years ago they transitioned their internal QA teams to development-role teams. Unless you’re referring to some particular XBOX gaming studio. For Windows there was some layoffs at the time of QA-to-dev transition, but they didn’t “fire all their in-house QA”. They just moved them to the current “industry standard” dev-qa-ops model, unfortunately.
I mean, it's true that the eliminated the engineer in test position. They didn't necessarily fire all the people in that position, because probably many of them switched to a new role.
But keeping the people and giving them a new role means the old role isn't being done, and I don't think it's done good things for product quality. The old way was expensive, and far from industry norms, but the product was more reliable.
Actually, for the Operating Systems Group specifically, it was reportedly a straight up layoff of all testers. See https://www.zdnet.com/article/beyond-12500-former-nokia-empl... Other divisions reportedly transitioned their testers to developers, which didn't work out well (big surprise) and resulted in heavy attrition later.
Did you read the article you linked to? The article is from 2014 saying that Microsoft laid off ~12K after acquiring ~18k of Nokia employees 3 years prior. That left ~5k out of total ~135k, at the time, unaccounted for per Microsoft announcement. Microsoft had 1:1 dev:QA before that. Even if those ~5k were entirely from OSG QA (they weren’t), it wouldn’t come close to “fire all their in-house QA”
I'm interested in the future naming schemes they are no doubt focus grouping now. Will we see Windows Series X? Windows '25? There's no way MS would do a Win13.
Agree. The problem being whatever support they provide (which means the updates) must be good enough so that people want it. And IMO I don't see the support being a good thing in Win 11 or recent. I currently stay on Windows due to better program support (some services provide less to none for Linux) and gaming (online games using anti-cheat not working on Linux). Hope Linux within a decade would gain enough market share & support to solve this pain points.
That's not a system update, though; you do actually need to actively run a pretty specific command to get the systemd-tmpfiles behavior. Closest thing I can recall was the old Steam for Linux bug that accidentally tried to rm rf /.
The summary of issue is "systemd-tmpfiles" also creates/manages '/home' in some cases and if one tries to run that command with "--purge" and expect it to just clean-up all generated temp files, it will delete '/home' too!
Not looking to start a flamewar, but sometimes the criticisms of massive scope creep in systemd do feel justified.
yeah, I never upgrade anything in windows besides my web browser as that's the primary attack vector
oddly enough I have a computer at work which works better than all the rest of my colleagues who go through updates on a weekly basis which often break their machines...
This is just a symptom of “Looks good let’s ship it and monitor logs” practice that almost every company has moved to, or an issue with some external driver.
For the latter, it’s unfortunate but you’ll rarely find a company diss`ing another company. Especially so if they are a Fortune 500 companies or if they are a major B2B company. Microsoft is both a Fortune 500 (Fortune 2?), and a major B2B company. You’ll never find a major B2B company dissing their B customers.
In the Linux world, the first is usually addressed with a “oh! the config for that moved from X to Y. We didn’t write a migration script, but did you not follow our open source development mail list? Recreate your config files following this example. We’ll write an update to handle it” or “fuck Nvidia/Intel/AMD, they are fucking morons and fucked up their drivers. Fucking idiots”
It could also turn out into a business model in future wherein the "enterprise" feature (to delay these updates) will cost a premium subscription - similar to how Youtube releases a premium version which only stops their crappy ads.
I'm actually 100% fine with this business model, if it's cost acceptable. I use YouTube for an hour+ a day and YouTube music for hours, having access to both of those services ad-free is fine by me. Their service is 6-13usd/mo (depending on your location) and gives daily entertainment, while other streaming services want you to pay 10-30usd/mo for maybe 1-2hrs/week of content, for maybe the length of a single series season.
What annoys me is creators double-dipping and adding sponsor reads throughout their video, to the point I have to run an extension (SponsorBlock) to get passed those. Or them gating arbitrary application specific features (PIP being a big one) behind their premium service.
Youtube premium still shows ads, but because they’re not a specific kind of injected non-skippable ads that interrupts the content, it’s somehow not an “ad” because content creators can hide behind the term “sponsored content”. It’s an ad by another name.
Yes. And according to recent reports, Youtube is also testing something called "sponsor block"[1] for their subscription tier. The AI will try to intelligently detect and skip that sponsored content in a video.
The best part of that approach is that it is completely unable to deal with anything that isn't logged in the first place, such as bricking a device.[1]
Similarly, it can't collect good info on performance issues such as intermittent lockups or freezes.
NVIDIA drivers for example collect an absurd amount of telemetry, but to this day their drivers freeze my computer completely (can't even move my mouse!) for about 10 seconds a short while after logging in.
No log, no problem.
[1] Oh sure, you're about to say that they can log the "started patching" event and then monitor for "didn't come back up" event afterwards. They could, but they don't. For years I had issues with blue-screens after forced semi-annual upgrades, and Microsoft was blithely unaware.
Oh yeah. I absolutely agree. The answer from management (or even your direct dev manager) is “we should log more”. It absolutely doesn’t matter how much you’re logging because there is always gonna be a drill-down question you can’t answer. “log more” is almost always the answer. At my current work we’re at a place where every function call has boilerplate of
True, but how often do you get to change your company’s/team’s/product’s entire framework/language/library because you got an ask vs “ok, just bolt on this call whenever you’re doning I/O”
There is a metric-fuck ton of telemetry on a windows machine (really any sufficiently sophisticated software/device).
Not only do they correlate the update start event with the update end event, they also have a billion other metrics that nosedive if the machine can’t actually reboot.
That doesn’t make it any more acceptable, but it’s the truth. It’s totally possible the first metric report doesn’t show up for a day, too. So, it’s doubly not ok for them to push something so haphazardly.
Even numbered Trek Films/FreeBSD releases/Windows releases don't suck?
I don't think it quite works though, because NT4 and NT5 (2k) were both good, although NT4 was hard to use with many games and win2k was better but still had some issues. I didn't hate XP on release and people grew to like it. So that's kind of 3 hits in a row, but then vista bad, 7 good, 8.x bad, 10 good, 11 bad kind of fits the pattern.
Didn't 10 introduce dark patterns to make users create an online account, forced telemetry, forced updates that can lead to unexpected reboots? (though unexpected reboot only happened once to me many years ago, they seem to have fixed that)
Windows 8 had a controversial UI but was respectful of user choices.
I have a hard time imagining a new Windows version not getting worse on respecting user consent.
54 comments
[ 4.4 ms ] story [ 125 ms ] threadI guess this justifies my decision.
MS stopped caring about quality long ago. Now they seem to be experimenting with how abusive they can be towards users.
Didn’t they fire all their in-house QA ages ago?
No, they didn’t? Like most other companies 10 (or so) years ago they transitioned their internal QA teams to development-role teams. Unless you’re referring to some particular XBOX gaming studio. For Windows there was some layoffs at the time of QA-to-dev transition, but they didn’t “fire all their in-house QA”. They just moved them to the current “industry standard” dev-qa-ops model, unfortunately.
But keeping the people and giving them a new role means the old role isn't being done, and I don't think it's done good things for product quality. The old way was expensive, and far from industry norms, but the product was more reliable.
At least I'm glad they keep an isolated channel for these early deployments.
> Windows 10 will reach end of support on October 14, 2025. [1]
So we've got a year and change. And they may push it back, like they pushed back EoL when people didn't upgrade from XP or 7.
[1] https://learn.microsoft.com/en-us/lifecycle/products/windows...
Perhaps "supported" should really mean "controlled by Microsoft".
The summary of issue is "systemd-tmpfiles" also creates/manages '/home' in some cases and if one tries to run that command with "--purge" and expect it to just clean-up all generated temp files, it will delete '/home' too!
Not looking to start a flamewar, but sometimes the criticisms of massive scope creep in systemd do feel justified.
oddly enough I have a computer at work which works better than all the rest of my colleagues who go through updates on a weekly basis which often break their machines...
For the latter, it’s unfortunate but you’ll rarely find a company diss`ing another company. Especially so if they are a Fortune 500 companies or if they are a major B2B company. Microsoft is both a Fortune 500 (Fortune 2?), and a major B2B company. You’ll never find a major B2B company dissing their B customers.
In the Linux world, the first is usually addressed with a “oh! the config for that moved from X to Y. We didn’t write a migration script, but did you not follow our open source development mail list? Recreate your config files following this example. We’ll write an update to handle it” or “fuck Nvidia/Intel/AMD, they are fucking morons and fucked up their drivers. Fucking idiots”
What annoys me is creators double-dipping and adding sponsor reads throughout their video, to the point I have to run an extension (SponsorBlock) to get passed those. Or them gating arbitrary application specific features (PIP being a big one) behind their premium service.
[1]: https://www.youtube.com/watch?v=4OX-OthHWAA
The best part of that approach is that it is completely unable to deal with anything that isn't logged in the first place, such as bricking a device.[1]
Similarly, it can't collect good info on performance issues such as intermittent lockups or freezes.
NVIDIA drivers for example collect an absurd amount of telemetry, but to this day their drivers freeze my computer completely (can't even move my mouse!) for about 10 seconds a short while after logging in.
No log, no problem.
[1] Oh sure, you're about to say that they can log the "started patching" event and then monitor for "didn't come back up" event afterwards. They could, but they don't. For years I had issues with blue-screens after forced semi-annual upgrades, and Microsoft was blithely unaware.
There is a metric-fuck ton of telemetry on a windows machine (really any sufficiently sophisticated software/device).
Not only do they correlate the update start event with the update end event, they also have a billion other metrics that nosedive if the machine can’t actually reboot.
That doesn’t make it any more acceptable, but it’s the truth. It’s totally possible the first metric report doesn’t show up for a day, too. So, it’s doubly not ok for them to push something so haphazardly.
2000 take.
ME, skip.
XP, take.
Vista, skip.
7, take.
8, skip.
10, take.
11? Definitely skip.
NT 3.1 - Good(?).
NT 3.5 - Bad(?).
NT 4.0 - Good(?).
2000 - Bad(?!)
XP - Good(?).
Vista - Bad.
7 - Good.
8 - Bad.
8.1 - Good(???).
10 - Bad.
11 - Good(?).
And before anyone asks, yes this is a bad meme and doesn't even make any sense.
I don't think it quite works though, because NT4 and NT5 (2k) were both good, although NT4 was hard to use with many games and win2k was better but still had some issues. I didn't hate XP on release and people grew to like it. So that's kind of 3 hits in a row, but then vista bad, 7 good, 8.x bad, 10 good, 11 bad kind of fits the pattern.
On the consumer side, there was 95, then 98, then 98 SE (Second Edition), and then ME.
...and with the exception of ME, all of those were actual improvements.
Windows 8 had a controversial UI but was respectful of user choices.
I have a hard time imagining a new Windows version not getting worse on respecting user consent.
1. "If we can't stop Windows botnets, we will just create our own."
It'll be smooth as butter, right?
Right?