[webkit-reviews] review denied: [Bug 38630] VS2010 asserts a null iterator passed to std::copy in Vector::operator= : [Attachment 55545] Kent's proposal patch v2

[bugzilla-daemon at webkit.org]

Alexey Proskuryakov <ap at webkit.org> has denied Jocelyn Turcotte
<jocelyn.turcotte at nokia.com>'s request for review:
Bug 38630: VS2010 asserts a null iterator passed to std::copy in
Vector::operator=
https://bugs.webkit.org/show_bug.cgi?id=38630

Attachment 55545: Kent's proposal patch v2
https://bugs.webkit.org/attachment.cgi?id=55545&action=review

------- Additional Comments from Alexey Proskuryakov <ap at webkit.org>
> The standard says neither "The OutputIterator parameter must be
dereferenceable"
> nor "It may be not dereferenceable if [first, last) is empty."  So we can't
> assume the VS2010 behavior is a bug.

The standard doesn't need to explicitly mention everything it allows. If there
is no restriction, then it's allowed.

Please make this fix only for MSVC2010, file a bug with them, and reference
this bug in a comment in code, so that we could know when to remove the
workaround.

-	 if (size() > other.size())
+	 if (size() > other.size()) {
	     shrink(other.size());
-	 else if (other.size() > capacity()) {
+	     return *this;
+	 }

This doesn't look like it could possibly work. Did you run regression tests
with this patch? Or am I just confused?