Created attachment 243528 [details] reproducer.shar See the attached port. Errors: > internal/netxlite/getaddrinfo_cgo.go:203:7: could not determine kind of name for C.AF_INET > internal/netxlite/getaddrinfo_cgo.go:207:7: could not determine kind of name for C.AF_INET6 > internal/netxlite/getaddrinfo_cgo.go:141:20: could not determine kind of name for C.AF_UNSPEC > internal/netxlite/getaddrinfo_cgo.go:140:22: could not determine kind of name for C.SOCK_STREAM
(In reply to Yuri Victorovich from comment #0) diff updating the actual/current port, and poudriere build log available on https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=270020 linked this bug as blocker for that one, but missed adding the context.
The same errors on lang/go121. Is upstream aware of this?
The same errors on lang/go122. Is upstream aware of this?
See this patch in PR 270020, it fixes the build of net/ooni-probe-cli : https://bugs.freebsd.org/bugzilla/attachment.cgi?id=252287
(In reply to Rene Ladan from comment #4) Linked the wrong patch somehow, use this one: https://bz-attachments.freebsd.org/attachment.cgi?id=252320
I don't think this is a bug in Go itself, unlike what this PR suggests. If you look at the offending file you can see that it includes /usr/include/netdb.h (because we are not on _WIN32) but that file does not define the four C.AF_* and C.SOCK_STREAM values anywhere, also not via recursive includes. These values are defined in /usr/include/sys/socket.h, which is what my patch ends up including and which fixes the build of the port.
(In reply to Rene Ladan from comment #7) Upstream OONI confirms this is a bug on their side, it is not a bug in Go. See https://github.com/ooni/probe-cli/pull/1639#issuecomment-2272804013 Closing this PR.
Thank you for the explanation!