Port Socket::Address to win32
#10610
Conversation
straight-shoota
added
kind:feature
kind:refactor
platform:windows
| {% if flag?(:win32) %} | ||
| Socket.ip?("::1:2:3:4:5:6:7").should be_false | ||
| {% else %} | ||
| Socket.ip?("::1:2:3:4:5:6:7").should be_true | ||
| {% end %} |
There was a problem hiding this comment.
This is a bit weird. Do you know why this differs?
There was a problem hiding this comment.
No idea, but probably a glitch in the implementation.
There are actually a number of expectations commented out in that spec because they fail with some implementations. I just thought it would be better to test that expectation for POSIX implementations. We could potentially skip the win32 branch because it's faulty. But this at least documents a known failure.
And maybe we could do something about fixing such issues outselves at some point.
There was a problem hiding this comment.
This now returns true on Windows CI on master since https://github.com/crystal-lang/crystal/runs/5044032259?check_suite_focus=true, which may be due to windows-latest upgrading to Windows Server 2022.
There was a problem hiding this comment.
Oh that's great. However, now we must check for the windows version 🙈 Or just skip this spec on windows...
|
@straight-shoota Could you clarify which one I should review? This looks very similar to #10650 . Should one of them be closed? |
Adds bindings for win32 API and integrates them with
Socket::Address. The winsock2 implementation is pretty close to the BSD socket implementation used on POSIX systems, so there's not much more change necessary.The first commit moves common features from
socket.crto a new filecommon.crwhich allows to require it without the main socket code (this is essentially a reboot of #6717).Implements a fraction of #9544