On 05/16/2014 07:10 PM, Niall Douglas wrote:
Thoughts?
With over one hundred libraries, Boost is also facing a scalability problem. As nobody can be an expert on everything, it is only natural that some kind of specialization, and therefore fragmentation, will take place. This is already the case with the project-specific mailing-lists. ISO C++ is facing the same problem, and has introduced study groups to focus on specific domains. Maybe we should do something similar. What I am proposing is not the project-specific mailing-lists (although some of these could probably be adapted for the purpose), but rather domain-specific fora, where we can experiment with various topics related to the domain before proposing libraries for Boost. These fora should allow competing designs to develop before settling on specific ones (unlike the current project-specific mailing-lists that are build around a specific library.) One example could be network programming. Boost has a splendid socket API in Asio, but no libraries that utilizes this. A Boost network study group could look into higher level components, such as HTTP frameworks, device discovery (ZeroConf), messaging middleware, RPC frameworks, and so on. Christopher mentioned another viable domain, namely real-time / embedded systems. There has even been voices on the std-discussion mailing-list suggesting a C++ SG for embedded programming.