Emit errors or warning to user about recursion of display about rust HOT 6 OPEN

I could have sworn there was a similar report about using to_string within an impl Display for from maybe 2017/2018, but can't find it 😅

@workingjubilee in the wild I've seen that simple cases where the chain repeats after just one or two jumps represent the bulk of them, so beyond the obfuscation issues, I'm convincing myself that a very limited call-graph analysis to look for unconditional recursivity is warranted. It's funny that it's always involving Display or Debug though 😄

Edit: found it #45838

from rust.

there is no use case for using : as standalone formatter, and it is better the compiler to emit error to use : as standalone formatter because it is likely user want to use :? but miss ? by mistake,

Unlikely to work, there's probably a crate out there that uses it, so it can't be removed due to backward compatibility guarantees.

can be a warnings?

from rust.

Macros, like the one you call there, are pretty good at obscuring these impl details, but yeah we should probably warn.

from rust.

And regarding the 2nd suggestion, what is your view

there is no use case for using : as standalone formatter, and it is better the compiler to emit error to use : as standalone formatter because it is likely user want to use :? but miss ? by mistake,

from rust.

there is no use case for using : as standalone formatter, and it is better the compiler to emit error to use : as standalone formatter because it is likely user want to use :? but miss ? by mistake,

Unlikely to work, there's probably a crate out there that uses it, so it can't be removed due to backward compatibility guarantees.

from rust.

Yes, we can lint on that. In fact we already lint on a few bad format_args! inputs.

from rust.