Jay
Jay
Perhaps it's an issue about your setup or cargo itself, I can see jemalloc-sys does respect environment variable changes: ``` > env JEMALLOC_SYS_WITH_MALLOC_CONF="stats_print:false" cargo build Compiling tikv-jemalloc-sys v0.6.0+5.3.0-1-ge13ca993e8ccb9ba9847cc330696e02839f328f7 Finished `dev`...
Unfortunately no. DCO check failed due to dcoapp/app#114. And I don't have the permission to force merge it yet. /cc @siddontang
According to your previous analysis in #116, the failure is caused by `-Wall` flag, and the flag is removed by `build.warnings(false)`. So passing the cflags to build script should be...
> so should work no matter what people have in their compilers set. You can make one lib work with an incorrect configuration, but how do you make all libs...
> they ended up using wrong strerror_t signature which may cause unintended consequences Yes, this is issue jemallocator or jemalloc should take care of. An error should be reported instead...
Now with latest change, it should report failure during dev build on Nix with _FORTIFY_SOURCE being set.
A few lines were missing from the final configure file. And the detection description is updated to be readable.
> and i see warnings (promoted to errors) We don't enable warnings_into_errors flag, and according to https://docs.rs/cc/latest/cc/struct.Build.html#method.warnings_into_errors, it's disabled by default. Can you compile github.com/jemalloc/jemalloc successfully?
Can someone provide full build logs? Especially the file in `target/debug/build/tikv-jemalloc-sys-*/{output,stderr,out/build/config.log}`.
I can't reproduce it on an alpine image that has clang 19. I notice that all provided logs in the issue are related to nix. However, I have never used...