Closed kpcyrd closed 1 week ago
For what it's worth I've been happy with cargo-c and the upstream maintainers have been very helpful. In rustls-ffi I'm moving towards replacing the old Makefile
approach (as linked to from the current curl docs) and instead recommending cargo-c as the blessed path for building the library (static or dynamic).
I've installed cargo-c
locally and tried it out, seems like the only thing that is missing is adding a capi
feature, and perhaps some options to use the pre-existing header file, and... that's it?
Started #3787.
Is your feature request related to a problem? Please describe. I'm trying to get the FFI interface of hyper up to speed with rustls-ffi, for integration in an operating system.
Compare the current curl hyper documentation (which wants you to manually build and link cdylib objects) to the current curl rustls documentation (
that generates and installs a .so file, headers and and pkg-config integration for you@cpu has pointed out this document is not using cargo-c, the rustls-ffi pull request linked below is a better resource to look into).Most of the work to get rustls to this point was done in this pull request: https://github.com/rustls/rustls-ffi/pull/274
Describe the solution you'd like I'd like hyper to integrate with the pkg-config ecosystem that Linux distributions build upon, preferably through cargo-c.
Describe alternatives you've considered The curl hyper integration is currently planned for removal in January 2025.
Additional context There may be some additional context in https://github.com/rustls/rustls-ffi/issues/345.
There was development work done on cargo-c itself specifically for rustls-ffi that you may want to consider, in case you have concerns about dynamic linking and ABI stability: https://github.com/lu-zero/cargo-c/issues/345