[dmd-beta] dmd 2.051 beta

[Walter Bright]

Don Clugston wrote:
> It's not a DMD regression. With the DMD from 1.050 with the recent
> Phobos, shows the same behaviour.
> So, the compiler itself has not regressed.
> I think a minor change to Phobos has triggered an instance of bug 4269.
> I completely commented out std.process and replaced it with
>
> import std.array;
>
> void bugshell()
> {
>     auto a = appender!string();
>     char[] g;
> //    static assert (is(typeof(a.put(g))));
>     a.put(g);
> }
>
> and it continues to reproduce the bug. If you replace "char[]" with
> "string", the error goes away.
> I'm still working on tracking this down.
>
>   

It's also not a memory corruption problem. I can reproduce the same 
issue on Linux (by rewriting std.process.shell so the windows version 
compiles on Linux). I don't remember seeing a memory corruption problem 
have exactly the same result on Linux and Windows.