other annotations

[Michael P.]

:-)

On Thu, 28 Jan 2021 11:52:11 -0800
Michael Truog <mjtruog@REDACTED> wrote:

> I added a description at https://bugs.erlang.org/browse/ERL-373 for
> using pure/impure instead of spec for Erlang type specifications (to
> provide the concept of a function type). Adding it as something for

`-spec` is now both pure and unpure^{*1}, thus must be regarded as unpure,
thus equals a hypothetical `-unpure`.

Only `-pure` is needed in addition?


> dialyzer to evaluate would be helpful. I don't believe it is very
> important to make it easy to create your own guard functions, but this
> could be a step in that direction.

I reckon it would be of high(est?) importance to make sure
that the probability of purity of homebrewed guard functions
cannot be lower than that of the current ones ...

~M

--

Car and cider give me a pleasant mediterranean
feeling, whereas cdr makes me shudder.