Vladimír Štill

> @asl Any idea how to workaround this Boost C++20 bug? > > https://stackoverflow.com/a/67702145 > > It's the only thing blocking compilation for C++20 for our CI. Otherwise, #4663 is...

Apparently boost 1.74 is the one to fix this: https://www.boost.org/users/history/version_1_74_0.html > [Format](https://www.boost.org/libs/format/): > Correct allocator usage (fixes C++20 compilation). (Glen Fernandes) So that means the one in Ubuntu 22.04 is...

The reason this fails in the sanitizer build has actually nothing to do with clang and everything to do with GCC 10, which is installed in that image, but not...

Why is the `p4test` assumed to be always present? Or is it just that `p4test` does not have any dependencies that are also not dependencies of the "core" used by...

> We could make it optional too but then we only have `gtest`. There is already a `ENABLE_P4TEST` option. So presumably it can be disabled already. Of course, building without...

I don't have many strong opinions about naming style. Personally I find `_t` suffix annoying and prefer uppercase for types which matches with the proposed style. But mainly I am...

> > Do you prefer to enforce this (by a CI check)? > > We are still missing Clang-tidy support for this (#4254). Would the clang-tidy still run very slowly...

@fruffy I've tried this few weeks ago after the recent changes like removal of StringRef. Sadly it still did not work easily. The main problem is that there is a...

@fruffy I've pushed what I had in my local branch here. The old one is https://github.com/vlstill/p4c/tree/forward-cpp-friendly-old.

> Instead, one should resort to std::string and std::string_view for transient values that are not required to be saved forever. I agree there 100 %. We might need to provide...