Re: std::map and std::set based on AVL, not RB trees.

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

Paolo Carlini <pcarlini@suse.de> writes:

| Gabriel Dos Reis wrote:
| 
| >In this particular case, do we insist on implementing something that is
| >hardly justifiable, apart from a literal reading that does no service?
| >I think, the answer must be no.
| >  
| >
| We would be adding additional complexity for implementing a behavior which
| is not mandated anywhere. Personally, I consider this stance wrong.

I agree that implementing a behaviour that does nobody any service is
a wrong stance, and this is a precisely the issue here. As of the
argument that we would have a complexity not required by the standard,
I believe it is bogus because we started having TR1 long before it
gets approved and yet it adds a far higher level of complexity.  Why
did we do that?  Because TR1 is useful.  Do you have any technical
argument that you would advance in Lillehammer for keeping the (wrong)
semantics as it is?  That is, unless you have a serious argument to
block this issue, I don't believe it is reasonable to say it is
"open", therefore me must refrain from taking any action.  

| That said, of course you are a maintainer like me, and, if you really
| want to
| implement that, I will not even try to stop you. Actually, no more comments
| from me about this issue.

The way we used to work is not that if one is maintainer one is free
to implement whatever.  We use to work by consensus.

-- Gaby