Hi,
(adding in CC Francois too)
On 08/07/2012 11:43 PM, Ollie Wild wrote:
On Sat, Aug 4, 2012 at 10:34 AM, Paolo Carlini <paolo.carl...@oracle.com> wrote:
First, I think we should add libstdc++ in CC.
Thus, I would recommend people working in this area to begin with
unordered_map, because in that case emplace is already available, assuming
that's really the point (and therefore reverting the patch was a good idea,
luckily an existing testcase helped us)
At the same time an implementation of emplace is definitely welcome, in
any case.
I've attached a patch for unordered_map which solves the rvalue
reference problem. For efficiency, I've created a new
_M_emplace_bucket method rather than call emplace directly.
Patch looks already pretty good to me. A couple of comments:
- Did you consider whether _M_emplace_bucket could be used as an
implementation detail of_M_emplace? Or figure out in any case a
smart(er) refactoring to share code with the implementation of
_M_emplace. I didn't study in detail your code, but I think we can avoid
some redundancy.
- For consistency at least, I think we should get rid of the make_pair
in the other overload too.
I've verified all libstdc++ tests pass (sorry for the previous
oversight) and am running the full GCC test suite now. However, I'd
appreciate any feedback on whether this is a reasonable approach. STL
hacking is way outside my comfort zone. ;-)
If this looks good, I'll take a stab at std::map.
That would be great. We have a PR which remained open way too much time.
The challenge is figuring out a way to avoid writing too much code
similar to what we have already for the inserts. Honestly, I tried, a
while ago, and failed, didn't see an easy way, but in any case, it's
time to have this stuff implemented - I hear Marc ;) - one way or the
other, for 4.8.0
Thanks,
Paolo.