Joel falcou <joel.falcou <at> gmail.com> writes:
On 28/06/11 05:09, Denis Shevchenko wrote:
[…] Could you write about the subject areas and technical directions for libraries which would be *REALLY* useful in Boost today?
Thanks.
Well, I'll probably looks like a party crasher here but.
I think currently Boost dont require *more* library, it requires a consolidated base of more *stable* libraries.
[…]
What I will find worthy to be done is actively move toward C++11 and starts putting the pressure back on compilers and explore the new practice C++11 can bring or see how it will modify existing one
[…]
I think the time is right now to pause a bit, look at the road we walked and consolidate ourselves, both a community and as a software project before engaging into the next part of the journey C++11 just opened.
While I agree with you on the need to "consolidate ourselves" on *stable* (as in stabilizing for future standardization), I think that we should *also* be "engaging into the next part of the journey C++11 just opened" at the same time (if only to keep having libraries to stabilize), just the same as GNU/Linux distributions have stable and experimental release at the same time. I'd see boost as both a place to explore how to solve new areas of interest (the OP point of view) and a place to find a stable consensus on the state-of-the art (your answer as I understand it). FWIW, to answer the OP question, I'd like to see boost solve unicode processing in C++ (I find current state-of-the art a failure except for the work of Mathias Gaunard) and multithread friendly data structures (could persistent data structures be implemented with smart pointers memory managment ? I'd like to give it a try, but for now I'm moving to Clojure for this kind of code). Best Regards, B.