Hi, There is a patch in ticket #7330, that adds move semantics to Boost.Function. It looks pretty trivial, so if there will be no objections, I'll commit it to trunk at the end of next week. -- Best regards, Antony Polukhin
Hi,
There is a patch in ticket #7330, that adds move semantics to Boost.Function. It looks pretty trivial, so if there will be no objections, I'll commit it to trunk at the end of next week.
Move emulation should be enabled explicity on compilers that don't support rvalue references. This is the way its done with Boost.Unordered which requires defining the BOOST_UNORDERED_USE_MOVE macro. This is because Boost.Move can be break users code on older compilers. See issue #6167 and issue #6311.
Move emulation should be enabled explicity on compilers that don't support rvalue references. This is the way its done with Boost.Unordered which requires defining the BOOST_UNORDERED_USE_MOVE macro. This is because Boost.Move can be break users code on older compilers. See issue #6167 and issue #6311.
No move emulation used in patch #7330. Move semantics enabled only for compilers that really do support rvalues. There is too much workarounds in Boost.Function for old compilers, using Boost.Move will definitely break someones code. -- Best regards, Antony Polukhin
Committed it in revision r80552 about a week ago. All regression tests pass successfully. If there will be no objections, I`ll merge changes to release branch at the end of this week. -- Best regards, Antony Polukhin
participants (2)
-
Antony Polukhin -
paul Fultz