ASIO supports non-header-only to speed up compilation, Beast should follow this pattern.
Actions required:
*.hpp impl files should contain only template code*.ipp impl files should contain code which can be compiled in a separate TUGood to know
This issue has been open for a while with no activity, has it been resolved?
FYI this isnt about speed but about
supporting a use-case of compiling into shared libraries.
Also, afaik, this system in asio is more complicated because there's a great deal of OS abstraction (os specific code) in asio and also there's hidden helpers via static initializers that don't play nice in a shared library context on all platforms.
Given this, I believe the use case would simply become:
I guess what I'm saying is don't take asio's structure or compilation options as a standard to follow. It's very asio-specific and beast != asio.
Also I'd be interested to see the use case where someone is constantly editing and using beast and incurring a significant performance penalty. Recent compiler/toolchain benchmarks are pretty stellar and so are standard processors these days. :)
Yeah, I looked through the .ipp files and there isn't really much that would go into a translation unit.
This issue has been open for a while with no activity, has it been resolved?
It looks like this issue has either been abandoned or resolved so I will go ahead and close it. Feel free to re-open this issue if you feel it needs attention, or open new issues as needed. Thanks!
Most helpful comment
Yeah, I looked through the .ipp files and there isn't really much that would go into a translation unit.