RFC: Drop Windows platform support

[zeenix]

Given that:

• I've no Windows machine to test any changes to Windows code and rely purely on the CI machines.
• I have 0 interest in Windows platform.
• With any refactor of the base types, Windows-specific code requires changes.
• Windows code being very low-level, requiring unsafe and dealing with strange APIs.
• I have only known 1 person who needs zbus on Windows.
• With 5.0 dropping DBUS_COOKIE_SHA1 auth breaking the Windows support.

I think we need to consider removing this support.

Having said that, I am not expecting a lot of changes in the future requiring changes to Windows code and would be more than willing to keep the support if #723 is fixed before the next API break (no timeline or intentions for that at the moment).

[zeenix]

CC @elmarco

[elmarco]

Having a portable IPC & DBus is big plus. It allows qemu-rdp and qemu-display to work on Windows, ANONYMOUS works in that case. Please don't remove Windows support. It's one of the strong reason why I contributed to zbus in the first place and I keep using it. Removing SHA1 was already a big disappointment. I would rather have a fully compliant and portable zbus implementation rather than a strim-down version. So #723 isn't strictly required atm.

[zeenix]

I would rather have a fully compliant and portable zbus implementation rather than a strim-down version

cookie auth is not at all useful for Unix and dbus-broker doesn't even implement it. I don't think dropping of cookie auth should be viewed as going against compliance. The spec does not require it. The only reason for it to exist these days is Windows. So one thing is for sure: cookie auth is not going to be coming back.

Please don't remove Windows support.

If I don't have to touch the Windows code much, I will not remove it. This includes the CI, which has in the past given me a lot of headache.