Paul Fultz II wrote:
If boost ever wants to support the standard `make && make check && make install` with cmake then this will be a cycle.
OK. Why is having a cycle unacceptable? Ah, wait, I get it, you can't 'make install' before you 'make check', and you can't 'make check' before 'make install'-ing both libraries first. Persuading people to support this will be... difficult. The standard answer is that the tools need to be made smarter. bpm doesn't have a problem with cycles because it doesn't have a separate install step. Your 'make check' step would need to be able to find the downloaded dependency in-place, without relying on it being 'make install'-ed first. So when you do cget boostorg/config, it would need to download Config, Core, Assert, Predef, Type Traits, and whatever else is required, then 'make check' at the download location using the downloaded libraries. Which is probably out of scope for cget, as it's generic and this workflow is boost-specific. How does cget find the dependencies now by the way?