Björn Fahller

I think you may have a reason to file a bug report with clang-tidy. Look here: ```Cpp #include struct S { virtual void func(std::function) = 0; }; struct SV :...

The sad thing is that these functions don't have to be implemented at all. They're only ever used on `decltype()` context to disambiguate overloads, so only the signature is needed....

I generally agree, but suppressing a diagnostic is by definition hiding a bug. The question is if its a bug in your code or in the tool emitting the diagnostic,...

This one's harmless and won't clutter code. I actually quite like trailing return type. I think the reason Trompeloeil isn't written with it is just that I hadn't quite gotten...

I just started looking into this, and curiously the reduced example seems to work with fmt trunk, but crashes with 8.1. https://godbolt.org/z/xc6hWoY5d

I'm beginning to think `fmt::is_formattable` is broken in 8.1 and earlier. If I change the criteria to the result of trying to parse the formatting string for the type, then...

I think `fmt` otherwise concludes that `trompeloeil::re` is printable as a C-string, since it wants to instantiate the conversion to `const char*` operator, which triggers the infinite recursion.

I don't think I like the `operator const char*` idea. Some library will do a similar thing but with `std::string_view` instead, or something similar. I'm a bit curious about why...

I will need to digest the standard. It feels odd to me that a function definition is required, but I can't argue against it. Does this mean we should implement...

Hmmm. I have to think about this one. You may have uncovered a bug, but I'm not sure ;-) The difference is that some support functions are not tagged as...