Does This Proposal Need Higher Scrutiny?

[RandomDSdevel]

I know that adding named arguments to C++ would be a great thing to accomplish, but it has been tried before without success. I just wanted you guys to take a look at (and, hopefully, learn from) the earlier, failed proposal from 1992 (yes, it's that old!), other people's analyses of it here and here, and page 14 of the minutes of a WG21/X3J16 meeting during which this same proposal was discussed. The result of all of this discussion was, I presume, the 'Named Parameter Idiom' (Google it if you don't know what it is.) I would also like to note here that someone has written a rather kludgy, but still interesting, implementation of keyword arguments in C++11. Please look at this research material which I have compiled for you and learn from any mistakes which you find within it!

[theres-waldo]

Thanks for the links! I was aware of the 1992 paper and its analyses, but not of the minutes which provide a good summary for why the proposal was rejected.

[RandomDSdevel]

You're welcome! Did you look at the custom implementation of keyword arguments in C++11 code? I'm just starting to learn how to program and think that named arguments would be a great addition to C++, especially with respect to Objective-C compatibility, but don't exactly know if the solution provided in that link would give you guys any pointers on creating your own temporary implementation (it looks kind of messy and inscrutable to me, but, then again, I am, like I said, just starting to learn C++…)

[theres-waldo]

I looked briefly at the library implementation you linked to. Boost has something similar.

We're not looking to write a library emulation of named arguments; I don't believe one would do much in terms of informing the direction of our proposal. However, after an initial draft of the proposal has been presented to the standards committee and we have some amount of encouragement to move forward with it, we may attempt an actual implementation in clang.

[RandomDSdevel]

You're quite right in saying that emulating C++ support for named/keyword arguments/parameters probably wouldn't be a good idea. If you guys didn't already know how you were going to implement this feature in a compiler such as LLVM/Clang, then doing so might give you some information about how to do so, but you guys obviously know what you're doing here (I, as somebody who is just starting to learn C++, am actually starting to become a little lost!)