On Saturday, 28 December 2013 at 23:38:38 UTC, Ivan Kazmenko
wrote:
Ouch, is it an exception hard-coded into the language itself?
I thought it's just the nextPermutation's parameter type
restrictions which don't allow "char []"...
No, a "char[]" is just a "char[]" as far as the language is
concerned[1]. However, for the *phobos* range abstraction, (the
front/popFront/empty) primitives, a char[] is a range of dchars.
Once you've defined char[] as such a range, it affects all of
phobos.
It's critiqued every now and then, but overall, it has proven to
increase code correctness, at the cost (sometimes) of rejecting
code that might be incorrect. Long story short: If you want to
handle UTF, you have to handle it *explicitly*. *I* think it is a
good compromise.
[1] the *only* place (AFAIK) that dmd looks at a string as a
range of dchars is in a "foreach(dchar c; s)" loop. But even
then, if you don't specify "dchar", then c will default to char.
