1..2 in typespecs does not mean a range %Range{first: 1, last: 2}, rather
an integer from 1 to 2.

On Tue, Apr 13, 2021 at 6:13 PM Felipe Stival <[email protected]> wrote:

> Shouldn't we support custom step types (like we do for first and last)
> like in this diff
> <https://github.com/elixir-lang/elixir/compare/master...v0idpwn:patch-1>?
>
> Em terça-feira, 13 de abril de 2021 às 18:17:35 UTC+3, José Valim escreveu:
>
>> Given Dialyzer does not natively support stepped ranges, I don't think we
>> should emulate them and have 1..10//1 and 1..10//2 to two very distinct
>> things. :(
>>
>> On Tue, Apr 13, 2021 at 5:03 PM [email protected] <[email protected]>
>> wrote:
>>
>>> While working with the v1.12.0 rc0 release, I wrote the following
>>> typespec:
>>>
>>> @spec my_fun(1..10//2) :: boolean()
>>>
>>> which resulted in the following error:
>>>
>>> == Compilation error in file lib/foobar.ex ==
>>> ** (CompileError) lib/foobar.ex:6: type ..///3 undefined (no such type
>>> in Foobar)
>>>
>>> My confusion stemmed from my belief that ranges in typespecs were
>>> interpreted by Elixir and expanded at compile time (e.g. 1..10 to
>>> 1|2|3|4|5|...) since the syntax (first..last) matches the Elixir range
>>> syntax and Erlang has no native ranges. Of course, this was wrong because
>>> while Erlang doesn't have ranges, *Dialyzer* does.
>>>
>>> While I can't think of the last time I wanted the contract of my
>>> function to be "only accepts odd integers between 1 and 100", it also seems
>>> like a curious omission. From the perspective of someone with good
>>> knowledge of Elixir but limited knowledge of Erlang and Dialyzer, I had
>>> originally believed that this was some kind of mistake or oversight.
>>>
>>> There also might be a use-case in providing additional information to
>>> Dialyzer, such as:
>>> @spec odd?(1..10//2) :: true
>>> @spec odd?(2..10//2) :: false
>>>
>>> Would it be worth taking a type such as 1..10//2 and compiling it to
>>> 1|3|5|7|9 for Dialyzer?
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "elixir-lang-core" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to [email protected].
>>> To view this discussion on the web visit
>>> https://groups.google.com/d/msgid/elixir-lang-core/d9415490-4285-48fe-a863-4d778ccc962an%40googlegroups.com
>>> <https://groups.google.com/d/msgid/elixir-lang-core/d9415490-4285-48fe-a863-4d778ccc962an%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>>>
>> --
> You received this message because you are subscribed to the Google Groups
> "elixir-lang-core" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/elixir-lang-core/b164a0dc-572e-4ec6-b62b-7d2ddff33fa7n%40googlegroups.com
> <https://groups.google.com/d/msgid/elixir-lang-core/b164a0dc-572e-4ec6-b62b-7d2ddff33fa7n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"elixir-lang-core" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/elixir-lang-core/CAGnRm4LShG5tQZysaWzNU-kYYi5p4X0HA-gOtKYsfJHh07iGyQ%40mail.gmail.com.

Reply via email to