[Development] Setters: Clarifying the ownership

[Marco Bubke]

Using a standard from the Cpp Core Guidelines will have the advantage of better tooling support. And please don't use the preprocessor, it will down slow the refactoring or makes it impossible.

________________________________
From: Development <development-bounces+marco.bubke=qt.io at qt-project.org> on behalf of Eric Lemanisser <eric.lemanissier at gmail.com>
Sent: Saturday, January 20, 2018 9:17:43 AM
To: Thiago Macieira
Cc: development at qt-project.org
Subject: Re: [Development] Setters: Clarifying the ownership

Please use an already existing solution to this problem : https://github.com/isocpp/CppCoreGuidelines/blob/master/CppCoreGuidelines.md#Ri-raw
Errors will be caught by the compiler in case of std::unique_ptr, and by static analysers (like clang-tidy) for gsl::owner, without run-time cost.
There was already an attempt to use gsl::owner https://codereview.qt-project.org/#/c/178107/ but it did not go forward, because of lack of time ?

Le ven. 19 janv. 2018 à 19:33, Thiago Macieira <thiago.macieira at intel.com<mailto:thiago.macieira at intel.com>> a écrit :
On Friday, 19 January 2018 09:26:10 PST Edward Welbourne wrote:
> Jaroslaw Kobus (19 January 2018 17:09)
>
> > "give" may be confused with "get", which is usually an accessor. I may
> > also think "Am I giving (to QCoreApplication)" or "The
> > QCoreApplication is giving (me)". Maybe it is just a matter of the
> > other verb? Absorb, hand over, hand on, suck in, swallow...
>
> However, we have plenty of take functions, where the caller takes
> ownership from the object on which the method is called; so it makes
> sense that a give function would be the caller giving ownership to the
> object on wich the method is called.
>
> Thus we'd keep child.setParent(newParent), since the child doesn't take
> ownership of the parent; but QMainWindow's setCentralWidget() would
> become giveCentralWidget(), matching its takeCentralWidget().  This
> would save the search for its doc, to find that it does indeed take
> ownership.
>
> The signature is, in any case, always sufficient to make clear that a
> give()r isn't a get()ter.

Let's stop the discussion about method *naming* right here. We're not going to
change hundreds of getters and setters now or even in Qt 6.

Let's instead find a solution that either uses macros or uses simple binary-
compatible pointer wrappers like GST.

And be careful with template functions. Changing from T to Something<T> may
change what gets deduced.

--
Thiago Macieira - thiago.macieira (AT) intel.com<http://intel.com>
  Software Architect - Intel Open Source Technology Center



_______________________________________________
Development mailing list
Development at qt-project.org<mailto:Development at qt-project.org>
http://lists.qt-project.org/mailman/listinfo/development
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/development/attachments/20180122/57eec774/attachment.html>