DavidSpickett wrote:

I will recuse myself from reviewing since I had input to the patch already, but 
want to highlight this that I said earlier:

> In theory someone could be relying on there being another clang-tblgen in 
> bin, maybe for installs, but that seems like a thing we can afford to break. 
> They were building clang-tblgen anyway, so they could stop setting 
> CLANG_TABLEGEN_EXE, and use the newly built one instead.

This is the main risk I see with doing this, and my guess of its impact may be 
totally wrong.

@rossburton could you add to the PR description the justification for the 
change, and the use case you want it for? Which I think is roughly - don't 
build clang-tblgen twice - it'll be used in some sort of build chain in Yocto.

https://github.com/llvm/llvm-project/pull/161952
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to