Nope, it doesn't just work. If anyone wants to look into this in more detail, feel free to do so.

@Al2Klimov Your recent activity in #9747 and #10199 seems to be related to me opening this PR (at least you added references to this PR). Can you please describe what your plan with that, just assigning yourself to PRs opened by someone else and reopening PRs and requesting review without any comment isn't too helpful. Do you want to point that there is already some work I could use as a basis? Do you want to point out that your PRs would better solve this (then you should say why windows-2022 + additionally installing another version of Visual Studio would be preferable to windows-2025 and just using the Visual Studio version already preinstalled on the runner)?

I didn't even look up what's preinstalled where. I just saw what you pushed and remembered that those stuff already exists:

• Bump the OS 52d9254? Already done: GHA: Windows: use Server 2022, not 2019 #10199
• Bump the whole dev env 8054fd5? Already done: Windows: upgrade to VS 2022 #9747

But, to answer your questions:

First of all, I stick to what I said in #9747/#10199: as long as nothing is EOL/broken, I'd do absolutely nothing.

But now you say (OP) that GHA will break the OS. If they actually do this, I'd consider #10199.

why windows-2022 + additionally installing another version of Visual Studio would be preferable to windows-2025 and just using the Visual Studio version already preinstalled

#10199 is already done and doesn't require us to overhaul the whole dev env incl. build server. It's admittedly not fast, but our Linux GHA aren't either and Windows is only 2 jobs. The more narrow a PR's scope (#10199), the better.

Do you want to point that there is already some work I could use as a basis?

Given I've already done the work, yes, if plain #10199 is not enough for you.

So, please let me know whether #10199 is fine.

• Just saw Release v2.15.0 #10471. If you really want Windows: upgrade to VS 2022 #9747, a .0 version is the-opportunityTM.

If you really want GHA: Windows: upgrade to VS 2022 #9747, a .0 version is the-opportunityTM.

I wouldn't say this is tied that closely to our release schedule. For one, if we go for Visual Studio 2022, we probably want to switch our builds for our support branches to that as well so that we don't have to maintain separate build toolchains. Also, with that statement

First of all, I stick to what I said in #9747/#10199: as long as nothing is EOL/broken, I'd do absolutely nothing.

That will most likely not align with our release schedule when it happens. So then we'd have to build whatever would be the next version then with the new toolchain.

#10199 is already done and doesn't require us to overhaul the whole dev env incl. build server.

That sounds quite dramatic. I doubt it's more effort to upgrade Visual Studio there than it's to downgrade it on the GitHub runners. Eventually, we have to switch to a newer version anyways, so I'd rather put work into that.

Then there's the question of the Windows version. Given that both GitHub's windows-2022 and windows-2025 runner images currently install the very same version of Visual Studio (17.13.35825.156 at the time of writing), I expect little to no difference between the two for our purposes. windows-2025 should have the advantage that it should have a longer remaining lifespan, hence I started trying that version with this PR, but to be fair, I haven't looked into this very much, for example I don't know when the next major version of Visual Studio is supposed to be released and which runner images would be updated to that hypothetical new version by GitHub.

So in summary, I prefer going to Visual Studio 2022. And if there was a problem with that, then this should be looked into instead of being ignored.