Comparison of Sepples and D

D:

no more preprocessor
no more headers
faster compilation
order of declaration is not important, no need to predeclare
a name conflict when importing modules with the same identifier trigger an error. Impossible to use the wrong symbol unknowingly.
no more implicit conversion of arrays to pointers
++pre and post-increment++ have been fixed
ranges are simpler to implement than iterators
the D STL is actually readable
move and copy semantics radically simplified
array slices are supported at the langage levels (pointer carrying their length), more bounds-checking ensues
C++ code is unittested at great pains : unit tests have been made a builtin
documentation comments are builtin to avoid the need for an external tool that inevitably no-one uses
D has a package manager, C++ has none that is popular in its community
stack variables / members are initialized by default. This fact alone would have saved me months of debugging C++ code
lexing stage separated from parsing stage separated from semantics
easier and more powerful templates that allow the average programmer to create meta-programs routinely
productivity generally increased by the GC
saner operator overloading

Sepples:

C++ seems a bit more thorough and specified
C++ story for resources is better

D:
Code slower than JavaScript(DMD is 3x slower than C++,v8 is 1.5-2.0x slower)
Stop-the-world Garbage Collector
Bugs unfixed for years, horrible sepples codebase.
Libraries in state of flux, cannot be relied onto.
Proprietary backend(DMD) of reference implementation http://stackoverflow.com/questions/15188957/does-d-language-have-any-proprietary-baggage
Windowsx64 compilation requires microsoft's linker(DMD)
All features are implemented in DMD, then ported to LDC and GDC which both have small teams. All effort for these years goes to DMD and the compiler remains as shitty and buggy as ever.
http://stackoverflow.com/questions/28922323/improving-line-wise-i-o-operations-in-d
Features change rapidly, new features cannot be relied on.