readln with buffer fails

[dcrepid via Digitalmars-d-learn]

On Wednesday, 29 October 2014 at 23:28:07 UTC, Justin Whear wrote:
> Part of what readln does is *modify* the slice itself, not just 
> the
> pointed-to
> characters. In particular it alters the length member so that 
> you know
> how much
> input was actually read.  This is also why the rvalue reference 
> shouldn't
> work.
> Remember, D chose not to repeat C's mistake of relying on null
> terminators.

Nice, thanks for that. I wasn't aware the .length member was 
changed, but I just verified it myself by surrounding the call 
with some debug output. Sure enough, its length is 4096 before 
the call, and a different length after (depending on what was 
input).