(DO NOT MERGE) fix: proc_macro_srv: temporary abi fix (rust-lang/rust#84717) #9047
Conversation
|
Canceled. |
|
bors r+ |
9047: fix: proc_macro_srv: temporary abi fix (rust-lang/rust#84717) r=lnicola a=lnicola bors r+ 9048: Add some lint completion tests r=Veykril a=Veykril bors r+ Co-authored-by: cynecx <[email protected]> Co-authored-by: Lukas Wirth <[email protected]>
|
Build failed (retrying...): |
9047: fix: proc_macro_srv: temporary abi fix (rust-lang/rust#84717) r=lnicola a=lnicola bors r+ Co-authored-by: cynecx <[email protected]>
|
Build failed: |
|
I think we need to wait for the |
lnicola
changed the title
|
We might need this too: rust-lang/rust@92b2894 |
|
Does this PR solve the issue yet? I tried manually merging this PR to use it, but it still resulted in the proc macro server crashing. Is there anything to make this work, at least on my own machine? |
|
We also need at least #9047 (comment). |
|
But that seems to already be a part of latest nightly? I'm trying to use the latest nightly as of right now, and thought merging this PR into master on my PC would work but apparently it doesn't. |
|
It's in the latest nightly compiler, but needs to be ported to that rust-analyzer PR. |
|
I made this commit on top of @lnicola 's adaptations to be able to use rust-analyzer with the latest nightly, so putting this link here in case it may save anyone a bit of time. |
9550: Proc macro multi abi proof of concept r=matklad a=alexjg #8925 was irritating me so I thought I would have a bash at fixing it. What I've done here is copy the `crates/proc_macro_srv/src/proc_macro` code (which is copied from `<RUST>/library/proc_macro`) to `crates/proc_macro_srv/src/proc_macro_nightly` and the modified the nightly version to include the changes from #9047 and https://github.com/yotamofek/rust-analyzer/commit/aeb7b183a2302af968f1110fb3cbd00c76a34974 This gives us the code to support both stable and nightly ABIs. Then we use the `proc_macro_api::version::read_dylib_info` to determine which version of the ABI to load when creating a `ProcMacroLibraryLibLoading` (which is now an enum). This seems to work for me. The code could be cleaned up but I wanted to see if the approach makes sense before I spend more time on it. I've split the change into two commits, the first is just copying and modifying the `proc_macro` crate, the second contains most of the interesting work around figuring out which ABI to use. Co-authored-by: Alex Good <[email protected]> Co-authored-by: alexjg <[email protected]>
|
Superseded by #9550 |
bors r+