#![no_std] testing

[workingjubilee]

It would probably be very valuable if some method could be developed that allows for true testing of actual #![no_std] Rust objects, even if it requires some additional work to set up, special harness support, whatever. Does the team have something in mind for this?

[epage]

As a team, we have committed to looking at json output as a way of improving test output by integrating test reporting into cargo test.. We haven't set specific priorities past that. I know some want to work to resolve problems with doctests, including their performance being so bad that some people avoid them. For myself, I want to look into making custom test harnesses first class with an extensible harness that means I can stop writing custom harnesses for so many of my projects.

I do not know of anyone on the team for whom this is a high enough priority at this stage. I do know that there are other interesting test problems related to this, like running tests in foreign environments (e.g. cargo test running tests in the Linux kernel)

[calebcartwright]

I think Ed said it very well. We'll keep this open as something on the backlog to discuss at some point, but I don't anticipate that this will get much attention from us for the foreseeable future

[jdonszelmann]

#3 helps (by creating a basis for supporting custom test harnesses)