[Development] Module maintainers: QT_NO_CONTEXTLESS_CONNECT in your modules

[Ivan Solovev]

Hi,

> Second, it's also easy to forget that since there's no receiver/context,
> the connection is always forced to be *direct*, which complicates things
> if multiple threads are involved.

I'm not a module maintainer, but I remember fixing a bug in Qt Bluetooth which was caused by exactly this issue.
And I must admit that it was rather difficult to detect it.

So, I think that this macro is useful.

Best regards,

------------------------------

Ivan Solovev
Senior Software Engineer

The Qt Company GmbH
Erich-Thilo-Str. 10
12489 Berlin, Germany
ivan.solovev at qt.io
www.qt.io

Geschäftsführer: Mika Pälsi,
Juha Varelius, Jouni Lintunen
Sitz der Gesellschaft: Berlin,
Registergericht: Amtsgericht
Charlottenburg, HRB 144331 B
________________________________
From: Development <development-bounces at qt-project.org> on behalf of Giuseppe D'Angelo via Development <development at qt-project.org>
Sent: Monday, July 10, 2023 6:02 PM
To: development at qt-project.org <development at qt-project.org>
Subject: [Development] Module maintainers: QT_NO_CONTEXTLESS_CONNECT in your modules

Hi,

https://codereview.qt-project.org/c/qt/qtbase/+/487560 introduces
QT_NO_CONTEXTLESS_CONNECT , a macro that disables the 3-arguments
connect -- in other words, it disables the

>   QObject::connect(sender, signal, functor)

overload, leaving only 4/5 argument(s) overloads

> QObject::connect(sender, signal, receiver/context, functor/slot, (type))


as well as the string-based connect().


The reason for NOT using the 3-args overload is that it is error prone.

For starters, it makes it hard to reason about the lifetime of such a
connection. It makes it very easy to connect to lambdas that capture
some local state in the receiver, but when the receiver is destroyed,
the connection isn't automatically disconnected (therefore, if the
signal is emitted, the program will crash). Fixing this may or may not
be straightforward, depending on how much state is captured.

Second, it's also easy to forget that since there's no receiver/context,
the connection is always forced to be *direct*, which complicates things
if multiple threads are involved.

--

I'm about to enable the macro for (most) of QtBase:

https://codereview.qt-project.org/c/qt/qtbase/+/489232

I've also done the exercise of enabling it for qtdeclarative (patches
pending) and qttools (merged), but there's simply too much code out
there that *still* uses it. As I said above, fixes are not always
obvious (sometimes it's very unclear who the "receiver" is).

--

Since this is ultimately a code style issue, I'd like to leave the
decision with the module maintainers regarding whether or not enable
QT_NO_CONTEXTLESS_CONNECT for their own modules. Ideally, however, we
should enable it in headersclean, so please make at least those not
contain 3-args connect().


Opinions? Objections?


Thank you,
--
Giuseppe D'Angelo | giuseppe.dangelo at kdab.com | Senior Software Engineer
KDAB (France) S.A.S., a KDAB Group company
Tel. France +33 (0)4 90 84 08 53, http://www.kdab.com
KDAB - The Qt, C++ and OpenGL Experts
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/development/attachments/20230710/93407f72/attachment.htm>