On Mon, Aug 22, 2011 at 03:35:04PM +0200, Olaf van der Spek wrote:
On Mon, Aug 22, 2011 at 3:21 PM, Mathias Gaunard <mathias.gaunard@ens-lyon.org> wrote:
Would something like this be possible to add to Boost.Containers, considering that going through the typical standard routes would have a latency of a couple years at best? I'm told that the Boost.Intrusive containers already supports this.
That would make Boost.Containers not standard-conforming, unless the function was given a different name.
Does the standard require exact function signatures?
Good question. On one hand, in the MS SC++L there's some variation in the set of insert overloads I believe. I could guess that one reason would be that code that relies on that there would be only one overload would suddenly cause unresolved-overload errors (use case: bind(&map<K,V>::find, this, ..) I was not quite aware that Containers was intended as a 1:1 implementation of the 11 containers, probably fooled by the set of non-standard containers bundled in the library. -- Lars Viklund | zao@acc.umu.se