Roadmap for CPP support

[fliiiix]

I think it would be good to think about the lowest cpp version this project targets.

See also comment by @dbaileychess #8858 (comment)

Are there reasons to stay on C++11? When should this be updated?

[jtdavis777]

To further complicate things ( ;) ) we should talk about this in two parts:

1. flatc - the command line tool
2. flatbuffers - the generated code and static library

These are currently very intertwined concepts, so disambiguating them may be difficult, but I do think they can have different support standards.

[jtdavis777]

Also see issue #8857

[jtdavis777]

CC @aardappel and @dbaileychess for comment

[aardappel]

This project in the past has tried to go more out of our way to support older compilers and older C++ versions as much as we can, for the simple reason that this a low level foundational technology, that is used often deep down in software stacks, with build systems that are not frequently updated.

The motivation to use newer C++ often comes from programmers wishing convenience in working on the FlatBuffers implementation, but usually does little to change things for the "end user" (those calling the generated code), since in the end this is meant to be rather simple functionality (looking up a field over an offset indirection).

We've of course have had a C++17 flag for the longest time to hide newer features behind.

The big issue is testing, where it may be hard to run CI on something as far back as C++11 and the compilers that used to be at that level. Still, I personally don't feel that's a good reason to start using newer C++ everywhere out of convenience. If someone comes across something that is broken on an older compiler, and has a PR to fix it, we should still want such changes.

While the core library code and generated code is more important to stay compatible with older systems than flatc, many build systems will build and run flatc as part of the same build system that is meant to compile the output, so I don't see a lot of benefit in terms of trying to separate the two.