[Interest] QVariantMap: constEnd() not equal to end()?

[André Pönitz]

On Wed, Apr 01, 2015 at 10:35:20AM +1100, Hamish Moffatt wrote:
> On 25/03/15 05:16, Thiago Macieira wrote:
> > On Tuesday 24 March 2015 17:29:39 Gunnar Roth wrote:
> >> Does anybody know why QT_STRICT_ITERATORS is not standard?
> > First, because it breaks existing, legitimate code that mixes const and non-
> > const iterators properly (yes, there are legitimate ways to do it).
> >
> > Second, because it's binary incompatible in QVector. The iterator for non-
> > strict QVector is a pointer; in strict mode, it's a class. So if you have the
> > iterator in your function parameters or template parameters, it will not be
> > compatible if you turn it on or off.
> >
> > However, I do recommend you turn it on in your own code, along with
> > QT_NO_CAST_FROM_ASCII and QT_NO_CAST_TO_ASCII.
> >
> 
> Thanks for the tip. I just enabled QT_STRICT_ITERATORS and fixed a few 
> bugs. I was bitten by comparison a const_iterator with a non-const end 
> after upgrading to Qt 5.4 (the same code worked in 4.8).
> 
> What's the benefit of QT_NO_CAST_FROM_ASCII though? I tried, but gave up 
> after I saw the first million compile errors.

Note there's also QT_RESTRICTED_CAST_FROM_ASCII that catches
less errors but is also less intrusive on your sources.

Given a choice, I'd prefer QT_NO_CAST_FROM_ASCII, though.

Andre'