unnecessary OS redundancy in druntime

[Joakim via Digitalmars-d]

On Saturday, 13 December 2014 at 15:58:29 UTC, Joakim wrote:
> When should it be redundant, for every single Posix declaration 
> in druntime?  That's far from the case now.  When the authors 
> think a handful of declarations might vary on some future 
> unspecified platform?  Each of the linked examples are 
> separated, yet none of them vary on any of the 4-5 supported 
> Posix platforms.  In fact, I just checked and none of them are 
> used anywhere in druntime and phobos, with only IPPROTO_RAW 
> used to define another unused enum in phobos.
>
> My guess is that you put each of these inside a version(linux) 
> block just to be safe when you first started druntime, and 
> porters have since been blindly adding other OSs to the list, 
> despite the separation serving no real purpose.  I have no 
> problem with having version(OS) blocks where there are actual 
> differences between the OSs, just not these cases where all the 
> OSs happen to be the same.
>
> I'm suggesting cleaning up this needless redundancy, to make it 
> easier for future porters.

Since I didn't get a response to this, I've submitted a 
work-in-progress pull with the suggested cleanups, along with 
others talked about in this thread:

https://github.com/D-Programming-Language/druntime/pull/1069

Any further discussion can take place on the PR thread.