Comments on: The C++ Metaclasses Proposal in Less Than 5 Minutes

By: Bartlomiej Filipek

Bartlomiej Filipek — Fri, 09 Mar 2018 09:33:00 +0000

lol “Runs Herb’s Examples” 🙂 Thanks!
I though the early implementation will be from MSVC… but probably it’s easier to extend clang to do prototypes.

By: Bartlomiej Filipek

Bartlomiej Filipek — Fri, 09 Mar 2018 09:32:00 +0000

In reply to Davidbrcz.

I know what you mean. C++ is already too complex, so metaclasses would only increase it…

on the other hand it will help in such places as QT – where you have its own metalanguage and you have to run MOC to generate the final code. (or use a recent solution without moc…)

so maybe metaclasses already try to simplify the existing (and non-standard) solutions.

By: Jonathan Boccara

Jonathan Boccara — Fri, 09 Mar 2018 08:44:00 +0000

In reply to Bartlomiej Filipek.

Thanks Bartek! It seems there is a clang-based prototype implem by Andrew Sutton available at https://cppx.godbolt.org/ that implements a number of features from the proposal

By: Davidbrcz

Davidbrcz — Fri, 09 Mar 2018 08:06:00 +0000

I have mixed feelings about the meta class proposal.
A C++ program is already hard to reason about and many things happen under the hood or behind our back. Adding a such a new powerful feature will break code locality (the type is a BLABLA, where the hell is defined BLABLA), that method is virtual because all BLABLA’s are..

Moreover, how does it fit with other C++ features ? Concept ? Auto ? Template ?

It may seem like a good idea at first glance but I really fear at the end it will only add fuel to the fire. You know, the road to hell is paved with good intentions …

By: Bartlomiej Filipek

Bartlomiej Filipek — Fri, 09 Mar 2018 06:50:00 +0000

great video! That’s a good summary of metaclasses.

Maybe I am wrong, but there’s probably some early implementation for meta classes… right? have you heard something about it?