"Ovanes Markarian"
To be honest, I tested the max_element with files (algorithms) from 1.34 and it did not work (map files did not update for 2 years). May be I forgot smth. The transformation of vector to map via fold worked. I can try to reverse the process and see if it works for map to vector. But I do not see any reason, to transform to vector first and afterwords iterate through vector again for finding the max element. As you can imagine, this will have impact on compilation speed. On the other hand, I can write my own traversal algorithm for maps. But it would be better to use some standard algorithms.
It sounds like you translplanted 1.34's MPL into your existing Boost installation? Remember I said that might not work.
Some probably weird question: If mpl::map fulfills the requirement of ForwardSequence, why wasn't it tested with for_each or max_element or may be some other useful algorithms?
Probably just an oversight. If you have some simple test cases we can add to the MPL test suite, please add them to the Boost SourceForge patch tracker. If they fail, maybe we'll fix a bug. -- Dave Abrahams Boost Consulting www.boost-consulting.com