On Wednesday, 19 July 2017 at 12:59:46 UTC, Dukc wrote:
I decided to try adding an overload to Zip which remembers it's
stopping policy at compile time. But I encountered another
strange design.
When stopping policy is set to requiring a same length, and it
is not so, it throws a regular exception. I think this is
inconsistent with rest of the language, because according to
TDPL exceptions are meant to intercept normal things the
programmer has no control over, Mainly I/O failures as I
understand it.
But here, if someone zips ranges of different length, it is
clearly a programmers error, not corrupted input. So shouldn't
it assert() the length (and be nothrow) instead?
And what would you think if I redesigned it so while I'm on it?
Nowadays that would probably be made an assertion and maybe put
in an in-contract, but zip is quite an old function so it doesn't
follow the more modern D conventions.
