[1.33.0] Unintentional dependency

I just realized that Boost.Parameter contains a dependency on boost/python/detail/is_xxx.hpp that I meant to fix before release by moving is_xxx.hpp into boost/detail. I'm certainly willing to wait until after 1.33.0 to fix this, but people complained bitterly about dependencies on boost/python/detail/indirect_traits.hpp before I did the same refactorization there. Doug, what do you think? -- Dave Abrahams Boost Consulting www.boost-consulting.com

On Aug 7, 2005, at 9:14 AM, David Abrahams wrote:
I just realized that Boost.Parameter contains a dependency on boost/python/detail/is_xxx.hpp that I meant to fix before release by moving is_xxx.hpp into boost/detail. I'm certainly willing to wait until after 1.33.0 to fix this, but people complained bitterly about dependencies on boost/python/detail/indirect_traits.hpp before I did the same refactorization there. Doug, what do you think?
I'm guessing that people won't be so bitter about dependencies in the Parameter lib because it is new this release. Wasn't the indirect_traits issue so problematic because it was introduced into several well-established libraries that didn't have dependencies on Python headers? I'm actually building the RC2 archives right now, so I'd prefer to leave this change post-1.33.0. Doug

Douglas Gregor <doug.gregor@gmail.com> writes:
On Aug 7, 2005, at 9:14 AM, David Abrahams wrote:
I just realized that Boost.Parameter contains a dependency on boost/python/detail/is_xxx.hpp that I meant to fix before release by moving is_xxx.hpp into boost/detail. I'm certainly willing to wait until after 1.33.0 to fix this, but people complained bitterly about dependencies on boost/python/detail/indirect_traits.hpp before I did the same refactorization there. Doug, what do you think?
I'm guessing that people won't be so bitter about dependencies in the Parameter lib because it is new this release. Wasn't the indirect_traits issue so problematic because it was introduced into several well-established libraries that didn't have dependencies on Python headers?
I don't remember. I think it was mainly the iterator library, which at the time was a total rewrite anyway. It was an issue for linux packagers who wanted to deliver Boost libraries in separate packages.
I'm actually building the RC2 archives right now, so I'd prefer to leave this change post-1.33.0.
Understandable. I think it's going to cause a problem, but sometimes that's the way the cookie crumbles. -- Dave Abrahams Boost Consulting www.boost-consulting.com
participants (2)
-
David Abrahams
-
Douglas Gregor