make ValueArg and MultiArg legal with g++11 c++20

[tomager]

Without this change, gcc 11.1.0 with c++20 selected returns "unqualified name expected before const". I think the existing code is not standard-legal but widely accepted, at least until c++20.

[DAarno]

This was fixed in b8e464f for 1.4 and 52a118c for 1.2. See http://sourceforge.net/p/tclap/bugs/33 for details and note that the master branch has been sunset upstream in favor of version branches (1.4 is current).

[tomager]

Oh sorry about that. It still appears on master which is confusing. Also appears in latest version captured by Conan center (albeit slightly stale on my end).

If master branch is not used any longer, it would be very convenient for the uninitiated to have a loud note about the branching scheme used at the top of README, since GitHub shows master by default. I can submit a PR for that if you like.

Thanks for your help!

[lapointexavier]

@DAarno If possible it would be great to get a new release for 1.2 to unblock this with GCC 11 🙏

[DAarno]

@lapointexavier, I'm not sure what you mean a "new release"? I think both 1.2 at HEAD and the latest source tarball (https://sourceforge.net/projects/tclap/files/) should compile with GCC 11. Let me know if I misunderstood something.

The problem here is that I don't control this git repo, My understanding is that it is just mirroring from the canonical sourceforge repo - where the master branch is deleted. Since it's deleted there isn't even a way to add a note.

I think the correct solution is for the owner (@john-peterson ?) of this repo to delete the master branch and set the default to 1.4, as is done in https://github.com/DAarno/tclap.

A workaround would be to pull from git://git.code.sf.net/p/tclap/code tclap-code (canonical) or https://github.com/DAarno/tclap.git

[lapointexavier]

@DAarno Sorry that wasn't clear. All I meant was a new release tag from 1.2's HEAD for convenience (Eg: v1.2.5). I assume this is what this person was referring to a few days back: https://sourceforge.net/p/tclap/bugs/33/#80b1 .

Otherwise yes the alternatives work for now. Thank you 👍

[DAarno]

Got'ya, I've cut a new 1.2.5 release that should build with GCC-11

[hjmjohnson]

It seems like this PR is no longer relevant, and could be deleted.