We could define a custom trait (see how hashbrown does this) to support it prior to the unstable trait landing. This would (at the very least) allow testing allocation fallibility[^1], if nothing else.
That said, this is a non-trivial amount of work that seems a bit pointless at the moment.
[^1]: Admittedly, the amount of code that it would allow testing is very small, and the code in question is very straightforward — just a few lines which do nothing but return errors to their caller.
We could define a custom trait (see how hashbrown does this) to support it prior to the unstable trait landing. This would (at the very least) allow testing allocation fallibility[^1], if nothing else.
That said, this is a non-trivial amount of work that seems a bit pointless at the moment.
[^1]: Admittedly, the amount of code that it would allow testing is very small, and the code in question is very straightforward — just a few lines which do nothing but return errors to their caller.