We currently pass a bunch of include paths into the compiler when compiling the C code that's part of rubyfmt. This is not that pretty, and requires adding extra code for every architecture/OS combination that we build rubyfmt on.
A better way is to ask Ruby what its "architecture" is, and use that information to construct the include paths that we need. This is slightly fragile, but is at least easy to make it fail loudly and early so people can figure out what's wrong if the formatting of this file changes.
froydnj
commented
1 year ago
We currently pass a bunch of include paths into the compiler when compiling the C code that's part of
rubyfmt. This is not that pretty, and requires adding extra code for every architecture/OS combination that we buildrubyfmton.A better way is to ask Ruby what its "architecture" is, and use that information to construct the include paths that we need. This is slightly fragile, but is at least easy to make it fail loudly and early so people can figure out what's wrong if the formatting of this file changes.
cc @ilyailya