On Monday 28 October 2024 08:39:59 Pacific Daylight Time Fabian Kosmale via Development wrote: > 1. Ensuring that CI gives us enough test coverage, which I suspect might be > solvable by configuring Qt to build examples and tests by default still > with 17.
I'll investigate, but I've never looked into how the CI tests compiling the examples. I also don't think it's worthwhile looking at the examples for this, for two reasons: 1) the headerchecking will be superior for the majority of issues, leaving only template expansions as problematic and our examples are not very template-heavy; 2) it may make sense to use new C++20 features in the examples. Compiling some of the tests in C++17 may make sense, but it might reduce coverage of the test itself. Do remember that there will be some platforms defaulting to C++17, so we'll keep support for that. > 2. Compile times, which I see getting worse when switching to 20, > and not improving until we can use "import std". And that's not probably > not viable even in C++23 builds unless mixing and matching of import and > include of SLT headers works reliably. I don't consider compiles slower a > blocker though, especially if compiling in 17 mode is still possible. Yeah, nothing we can do about that trend. > However, if it comes to ABI, I'm not convinced that we have a solution. As > long as e.g. libstdc++ declares its support experimental, are we really > willing to export C++20 symbols from Qt? If not, do we have strategy to > prevent this from happening? We don't have a solution for ABI-use right now, unless we can convince the libstdc++ maintainers to keep their current ABI for us. If all Jonathan wants is a beer count, I'll provide it :) Our current rules should be more than enough to prevent unintentional use of their ABI in ours. -- Thiago Macieira - thiago.macieira (AT) intel.com Principal Engineer - Intel DCAI Platform & System Engineering
smime.p7s
Description: S/MIME cryptographic signature
-- Development mailing list Development@qt-project.org https://lists.qt-project.org/listinfo/development
