These VPATH builds are supported natively by autotools, so we can just rely on them.
This also removes the leftover obj directory and provides an empty build
directory instead.
These VPATH builds are supported natively by autotools, so we can just rely on them.
This also removes the leftover obj directory and provides an empty build
directory instead.
These VPATH builds are supported natively by autotools, so we can
just rely on them.
This also removes the leftover `obj` directory.
What advantage does this gain? For coverage it seems fine to me, but 99.9% of users won't read the build instructions, and will just type ./configure && make like almost all other packages. Maybe then they'll open PRs adding obj to .gitignore, but in any case they'll be using something different from what gets tested.
For coverage, that's weird enough that RTFM is expected and even if people don't do whats expected it would impact far fewer people.
I think it's always a good idea to keep the tree as clean as possible. But that's my personal preference and yeah I agree, I guess we shouldn't recommend it to users if it's too uncommon and makes testing more complicated.
Maybe then they'll open PRs adding obj to .gitignore, but in any case they'll be using something different from what gets tested.
Not sure what you're referring to. If I'm not mistaken, the obj directory is currently unused, we should remove it in any case.
Ah, the project uses subdir-objects so indeed obj is unused. Sorry, I was confused by the fact that this PR removed it, as I assumed they it did so because they'd end up under the build directory. :)
Closing in favor of #941.