Socket::Constants incorrect on Windows

[stakach]

irb(main):001:0> require 'socket'
=> true
irb(main):002:0> Socket::Constants::AF_INET
=> 4
irb(main):003:0> Socket::Constants::AF_INET6
=> 33
irb(main):004:0> RUBY_DESCRIPTION
=> "jruby 1.7.10 (1.9.3p392) 2014-01-09 c4ecd6b on Java HotSpot(TM) Client VM 1.7.0_07-b10 [Windows 7-x86]"
irb(main):005:0>

According to MSDN values should be

• Socket::Constants::AF_INET == 2
• Socket::Constants::AF_INET6 == 23

[ahorek]

ping @enebo

jnr/jnr-constants#31

[enebo]

@ahorek I merged that I and I guess we will need to run a script to populate for next jnr-constant release but I see we hard-coded numbers in that commit above. Maybe we can get more updated for AF_INET and AF_INET6 are generated? Or is that not in jnr-constant?

[ahorek]

@enebo thanks
it would be great if you could review jnr-constants pull requests
especially these two should be ready
jnr/jnr-constants#30
jnr/jnr-constants#28

after that we have to run the generation script for all platforms and commit generated files separately. Sounds good?

[ahorek]

cotag/libuv@22e0ed6 no, this one isn't in jnr-constant repository. It was a workaround for https://github.com/cotag/libuv

jnr/jnr-constants#31 will fix it properly

[headius]

I've merged all mentioned PRs for jnr-constants. There's no changes when regenerating for MacOS.

[ahorek]

there should be changes for all supported platforms including macos.

I ran these commands on linux and windows

rake regen:xplatform
rake regen:platform
rake regen:fake

but something went wrong, do you have a tip how to fix these failures?

Lack of description for EAUTH expected not same
Incorrect integer value for EAUTH, expected:<0> but was:<20000>
Incorrect integer value for EAUTH, expected:<20000> but was:<20014>

https://github.com/jnr/jnr-constants/pulls/ahorek

[ahorek]

@headius it's ready to review. I fixed tests and added constants for freebsd, windows, linux and solaris, but I don't have any macos, aix and openbsd running.

[headius]

Hmm... I ran those commands too and didn't get anything new. I'll try again.

[headius]

Closing the loop... jnr/jnr-constants#41 and related PRs were needed to fix a bunch of decay in the generation of jnr-constants. Those are now merged, @ahorek is regenerating for their available platforms, and we should be good to go with a jnr-constants release for 9.2.1.

[headius]

Looks like some failures in Errno we'll have to look into.