Mac tests failed. One is related to travis-ci/travis-ci#6522 while I don't known why another fails. I don't have a working Mac so I can't test

I think this patch should probably be discussed on the list; the patch itself seems ok but the question is whether this is the desired direction. I don't actually know the rationale for the current approach.

Thanks. Is libffi-discuss the correct list for discussion about such a
change?

On Nov 15, 2016 03:32, "Tom Tromey" notifications@github.com wrote:

I think this patch should probably be discussed on the list; the patch
itself seems ok but the question is whether this is the desired direction.
I don't actually know the rationale for the current approach.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#288 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AB2RGQzldbBeH4BQpfn_tjUMU6h5KyjHks5q-Lc4gaJpZM4Kwqvm
.

Yes, libffi-discuss. Thanks.

Seems nobody on libffi-discuss is interested in this :/
https://sourceware.org/ml/libffi-discuss/2016/msg00053.html

Any thoughts on this? The change here is the correct way to handle header installation. What is controversial about it? Maybe those wishing to maintain the current behavior could advocate for it so we could understand the value.

There's nothing controversial. Instead, there are no real discussions about this. I consider libffi semi-dead - there's no development activity (mailing list posts, new commits) for almost two months! For those who need this patch, I'm afraid that the only way is maintaining a private copy of libffi and adding whatever necessary patches.

Yes, I was hoping for someone to reply on libffi-discuss to explain why the current setup is needed. But nobody did :(. I will see if I can find more information through some other channel. FWIW I'm generally in favor of making libffi's configury less weird; and making this change is one of the things I'd do.