I'm working at a small software shop, so my manager is open to suggestions for improvement. VC++ 2012 with c++11 support would be nice, but is probably a bit pricey. Instead, I want to convince him and other devs that we should start making use of the boost libraries. I have seen a fair number of C++ professionals swear by these libraries, and the functionality I have seen available when looking for solutions to various problems seems to be very useful.
So, the question: what libraries should I research and show to make the best impact in the shortest time? We use STL and win32 reasonably extensively (lots of wstring, etc), and a quite a few custom quick-make-it-work type of projects. Any suggestions?
Tell him that the people who maintain Boost are the same people who maintain the C++ standards. If you want something in C++, you first implement it as an approved Boost library. Almost all of the current C++11 updates came from the existing Boost libraries.
Also, it might be worth your time to go look over the boost.org website itself and peruse the libraries you think you will use.
Remind your boss that this is free and (relatively) easy to install and stupidly easy to use.
(The hardest part is installing everything into the correct spots.)
BTW, if you are using Unicode, install ICU also. (Do it first, since Boost locale, filesystem, regex, etc will use it as its backend.)
I can't imagine a C++ software shop that doesn't use boost, especially small one, not burdened with its own decades-old corporate libraries. The advantage of the small guys is agility and fast adoption.
Look at your existing code base, If a lot of it is indeed "quick make it work", then the more you have reliable and reusable components at your disposal, the faster you will make it work. Also, as it's been the best known general purpose library collection for over a decade, plenty of people are familiar with it, hence maintainability, quicker learning curve for the new guys.
Thanks for the ICU tip Duoas, appreciated! Looking for specifics to demonstrate...I think there is a time lib that replaces time_put, multi-reader single writer mutex, obviously hash_set...other good options?