Roland Schwarz wrote:
Peter Dimov wrote:
Maybe it's time to switch to HEAD?
I am not there yet. But if you think it is reasonable to postpone the boost.build v2 code, I can concentrate on getting the code to the state where it is possible to switch from thread_rewrite to HEAD. It should be much easier now since I know I can take Bills original code.
However I have some concern about the windows version which is done by Anthony Williams. His new code departs from the current code in larger areas from Bills original.
One of the reasons for switching to HEAD is that anybody who is concerned about the fate of the library can see what is being done. I realize that one could track a branch if one really wanted to, but this is just not what happens in practice. If the windows version conforms to the specification and passes its tests, there's nothing to be concerned about; if there are positive interface changes (to pick some arbitrary examples, a better call_once, a working and efficient read/write mutex, or a condition that conforms to N2094 if we decide to pursue this) for which the windows version is a proof of concept implementation, this is a good thing too. But it needs to be on HEAD and it needs to be regression tested. We already "lost" much of Bill's work because it was done on a branch. Branches are evil. Incremental refactorings that cause no unintended regressions are good. :-)