Subject: Re: [boost] Dependency report
From: Gavin Lambert (gavinl_at_[hidden])
Date: 2017-07-28 00:55:55
On 28/07/2017 02:53, Michel Morin wrote:
> After the move
> [Boost.Iterator module]
> Contains the following:
> - include/boost/iterator/next_prior.hpp
> * Contains the actual code for `next` and `prior`.
> - include/boost/utility/next_prior.hpp
> * Just includes <boost/iterator/next_prior.hpp>.
> [Boost.Utility module]
> Does not contain anything related to next_prior.hpp.
While this arrangement sort of makes sense for the purposes of cleaning
a dependency graph, doesn't this just proliferate the problem of header
files in the "wrong" directory for its library?
ie. shouldn't the forwarding header file still be owned by Boost.Utility?
Admittedly that doesn't fix the dependency graph, unless you make it
ignore that forwarding header ... which is probably only ok if you can
also guarantee that no in-Boost code actually *uses* that header,
otherwise it's still a real dependency.
But there's probably a few other compatibility/forwarding header files
that might fit that bill and should perhaps be similarly excluded, which
might mean the dependencies aren't quite as bad as they currently appear.
I guess it's a question of which problem is worse.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk