something like `ip_inputs:true` might work.
I usually update a container in the task, and return it, so it contains the
input + the output.
But if we had something like zip_inputs, then the task itself could just
return the calculated value, and it's up to the pipeline what to do with
them. I actually like it.
What would the output look like?
[{:ok, _input_, output}]
[{:error, _input_, :timeout}]
or something along the lines of
[{_input_, {:ok, output}]
[{_input_, {:error, :timeout}]
On Friday, 18 February 2022 at 11:08:25 UTC+1 José Valim wrote:
> Maybe we should have a zip_inputs: true or similar flag and we attach the
> input to all "ok" and "exit" tuples.
>
> On Fri, Feb 18, 2022 at 11:03 AM vtm <vtmil...@gmail.com> wrote:
>
>> Hi Jose and Juan.
>> I think there is a problem that you should have `order: true` to have a
>> correct result in Enum.zip
>> Also you might wanna have a big list and just use `big_list |>
>> Task.async_stream(params) |> Stream.filter(stream, &is_error/1) |>
>> Stream.map(&handle_error/1) |> Stream.run`
>> just to capture error with values and do some work
>>
>>
>> пт, 18 февр. 2022 г. в 12:48, José Valim <jose....@dashbit.co>:
>>
>>> Hi Juan,
>>>
>>> In your example a simple Enum.zip would suffice:
>>>
>>> list = [1, 2, 3, 4]
>>>
>>> list
>>> |> Task.async_stream(
>>> fn entry ->
>>> if entry == 3, do: :timer.sleep(2000)
>>> entry * entry
>>> end,
>>> timeout: 1000,
>>> on_timeout: :exit_task
>>> )
>>> |> Enum.zip(list)
>>>
>>> Can you provide an example where zipping would not be possible or too
>>> cumbersome?
>>>
>>> On Fri, Feb 18, 2022 at 10:38 AM Juan Peri <eternop...@gmail.com> wrote:
>>>
>>>> In these past few year I've found myself needing to execute several
>>>> async tasks (mostly around getting remote resources) in scripts, and
>>>> wanting to do something in the ones that fail
>>>> With the current implementation of async_stream, there is no way to
>>>> know which ones failed, as the output would look like the following:
>>>> ```
>>>> [1, 2, 3, 4]
>>>> |> Task.async_stream(
>>>> fn entry ->
>>>> if entry == 3, do: :timer.sleep(2000)
>>>> entry * entry
>>>> end,
>>>> timeout: 1000,
>>>> on_timeout: :exit_task
>>>> )
>>>> |> Enum.to_list()
>>>> [ok: 1, ok: 4, exit: :timeout, ok: 16]
>>>> ```
>>>> That would force me to get creative with stream Zip, or hand roll my
>>>> own solution.
>>>>
>>>> @vtm in the elixir slack pointed me to the commit
>>>> https://github.com/elixir-lang/elixir/commit/c94327cc4 in wich such
>>>> functionality was removed because of backwards compatibility.
>>>> And the commit that introduced it was
>>>> https://github.com/elixir-lang/elixir/commit/6c1fe08a676f1cacb7ee66dd56251b8a48a02d63#diff-0e81c1197153b352765d1d43ca57817901d19813a220e95a2cd2e70f3cfaa279R376
>>>>
>>>> when adding the :ordered parameter to Task.async_stream
>>>>
>>>> Would you accept a PR about exposing this functionality again, but
>>>> behind a opt-in new parameter that is false by default in order to keep
>>>> backwards compatibility.
>>>>
>>>> Using the previous example, it would look something like:
>>>> 1, 2, 3, 4]
>>>> |> Task.async_stream(
>>>> ...,
>>>> timeout: 1000,
>>>> on_timeout: :exit_task_with_value,
>>>> )
>>>> |> Enum.to_list()
>>>> [ok: 1, ok: 4, {:exit :timeout, 3}, ok: 16]
>>>> ```
>>>> I'm not sold in the flag :exit_task_with_value, it could be a separate
>>>> parameter as well, as long as it's opt_in
>>>>
>>>> Thanks
>>>> *_________________________________________________________**_**____*
>>>> *Juan* -
>>>> *No todo el oro reluce.....Ni todo errante anda perdido*
>>>> *¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯**¯**¯**¯¯¯¯*
>>>>
>>>> --
>>>> 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 elixir-lang-co...@googlegroups.com.
>>>> To view this discussion on the web visit
>>>> https://groups.google.com/d/msgid/elixir-lang-core/CAPx9ufN-kdNYuAJri_z9pxQ-0yopvSdNUVBLpZ1G6fTHPk1BDw%40mail.gmail.com
>>>>
>>>> <https://groups.google.com/d/msgid/elixir-lang-core/CAPx9ufN-kdNYuAJri_z9pxQ-0yopvSdNUVBLpZ1G6fTHPk1BDw%40mail.gmail.com?utm_medium=email&utm_source=footer>
>>>> .
>>>>
>>> --
>>> You received this message because you are subscribed to a topic in the
>>> Google Groups "elixir-lang-core" group.
>>> To unsubscribe from this topic, visit
>>> https://groups.google.com/d/topic/elixir-lang-core/XCQIKQc3KxM/unsubscribe
>>> .
>>> To unsubscribe from this group and all its topics, send an email to
>>> elixir-lang-co...@googlegroups.com.
>>> To view this discussion on the web visit
>>> https://groups.google.com/d/msgid/elixir-lang-core/CAGnRm4Li9mcnQ2mqN4xKb0qW9AX5uFi%2BXD77BDrqR-%3D0gW_8iA%40mail.gmail.com
>>>
>>> <https://groups.google.com/d/msgid/elixir-lang-core/CAGnRm4Li9mcnQ2mqN4xKb0qW9AX5uFi%2BXD77BDrqR-%3D0gW_8iA%40mail.gmail.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 elixir-lang-co...@googlegroups.com.
>>
> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/elixir-lang-core/CAKrSanzzRBYyFCM6T6dFRWoXVPBj-k2jbci2wfEsJa4bnDvL7g%40mail.gmail.com
>>
>> <https://groups.google.com/d/msgid/elixir-lang-core/CAKrSanzzRBYyFCM6T6dFRWoXVPBj-k2jbci2wfEsJa4bnDvL7g%40mail.gmail.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 elixir-lang-core+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elixir-lang-core/a9050152-816d-40c1-9845-4d4fdd831285n%40googlegroups.com.
