[Win32] long path name support [Bug #12551]

[GabrielNagy]

Implement long path support on Windows by applying Microsoft's recommended application manifest.

To make this work on both Visual C++ and MinGW, include the manifest as a resource when generating the resource files. This way it will be embedded into the executables/libraries generated by both compilers.

It's important for the manifest resource to have ID 1, otherwise GCC will embed a default manifest. Resource type 24 corresponds to manifest resources.1

Note that in addition to this, the user needs to have long paths enabled either by modifying the registry or by enabling a group policy.2

[GabrielNagy]

I tested this using the following steps:

1. Enable long paths either via Registry or Group Policy: https://docs.microsoft.com/en-us/windows/win32/fileio/maximum-file-path-limitation?tabs=cmd#enable-long-paths-in-windows-10-version-1607-and-later
2. Reboot (not sure if necessary)
3. Run a Ruby script located in a long path (over 260 chars) - this fails without the PR
4. Require a Ruby file located in a long path (over 260 chars) - this fails without the PR
5. Run Dir.mkdir with a long path - this fails without the PR

[GabrielNagy]

This can alternatively be done by adding the manifest resource to https://github.com/ruby/ruby/blob/master/win32/resource.rb, so this will also work with MinGW builds. I'll test some more.

[GabrielNagy]

This works for me with Visual Studio 2019 and GCC 10 (MinGW). I think AppVeyor compiles from a different directory so that's why it can't find the manifest, I'll try to see if I can get the basedir somehow.

[GabrielNagy]

Works! Nice!

There was a failing test when I first pushed the PR: https://github.com/ruby/ruby/runs/2591987561?check_suite_focus=true. It seems related but I think it confirms that long paths work:

  1) Failure:
TestProcess#test_spawn_too_long_path [D:/a/ruby/ruby/src/test/ruby/test_process.rb:1758]:
[ruby-core:34842].
[Errno::ENOENT, Errno::E2BIG] exception expected, not #<Errno::EINVAL: Invalid argument - echoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoechoecho>.

Should EINVAL be included in the allowed exceptions?

[michaeltlombardi]

@nobu what versions of ruby will this land in? 👀

[KostyaSha]

@larskanis @nobu what ruby versions will include this path fixes?

[larskanis]

I added the patch to all ruby branches of RubyInstaller here. That means that all versions of the next release series from 2.5 to 3.0 will include it. And it's already present in RubyInstaller-head.

[Inversion-des]

I just got this Errno::ENOENT problem in the development
and a small minute upgrade from v3.0.1 to v3.0.2 solved the problem — it was fantastic 😲!

Thank you so much! 👍❤