[Niall Douglas]
The VS2015 CTP6 changelog shows no bugs fixed in the compiler (https://support.microsoft.com/kb/2967191)? This is quite surprising. There are a number of compiler bugs yet, though it's night and day compared to VS2013.
Compiler fixes are being continuously checked in, but the people who write the changelogs don't always remember to ask the compiler devs for a list. Nor did they ask me for a list of STL fixes. I keep track of all STL fixes that are checked in, so I can write VCBlog posts about them (as I did for CTP1 [1], and will do again for RTM). But I don't immediately know what got into CTP6 - I'd need to ask somebody to verify what branch it was built from, or install it in a VM. I also keep track of STL features and compiler features (again for VCBlog posts) but not compiler fixes - there are just too many fixes and I can't accurately describe most of them. Approximately 80 STL bugs in our internal database (including Connect) have been fixed after CTP1. Notably, I overhauled <functional> to fix all known bugs, with the only limitation being Expression SFINAE. If the CTP6 branch is $/devdiv/Rel/D14Rel as the version number would seem to indicate, then it appears to contain both the overhaul and my later implementation of C++17 std::invoke(). This also means it contains fixes for several compiler bugs that I encountered along the way, including the interaction of PMDs with rvalues, and partial ordering among variadic templates. STL [1] http://blogs.msdn.com/b/vcblog/archive/2014/06/06/c-14-stl-features-fixes-an...