Note: I've only seen this message now, since I am several threads late
in the (date-ordered) queue of unread NG threads, and this message
appeared as a new thread.
On 06/10/2010 21:00, bearophile wrote:
Bruno Medeiros:
[...mumble mumble...]
I don't like this term "undefined behavior"
[...mumble mumble...]
I really don't care about words, and about C, I want signed/unsigned
compile-time/run-time overflow errors in D.
Bye,
bearophile
Like I mentioned afterwards, I think communication is important, so we
should strive to have a clear understanding of the terms we and other
people use.
But anyways, regarding this issue, I am satisfied. The D glossary and
TDPL have precisely defined "undefined behavior", which I didn't know
was the case.
Also, the related term "implementation defined", which some people in
the C world equivocate with "undefined behavior", has been used here in
D, but in also in a more accurate way, distinct from "undefined
behavior". So that's good.
--
Bruno Medeiros - Software Engineer
