I took a bit of a different approach: instead of spawning new processes
when all the conditiones match: I spawn them all as futures and they block
until their dependent futures return:
https://github.com/johnwahba/async_with
https://github.com/johnwahba/async_with/blob/master/test/async_with_test.exs
<- (has an expanded macro in the comments)
On Friday, September 8, 2017 at 2:12:47 PM UTC-7, Fernando Tapia Rico wrote:
>
> Thanks for the feedback Theron!
>
> Interesting...instead of using the dependency graph you are spawning new
> processes when all the conditions (the right vars) are matched. It's
> definitely an idea worth exploring.
>
> On Friday, September 8, 2017 at 9:52:00 PM UTC+2, Theron Boerner wrote:
>>
>> Hi Fernando,
>>
>> Great work on this implementation! However, there are a few edge cases
>> that you do not handle. E.g.
>>
>> async with a <- 1,
>> b <- :timer.sleep(1000),
>> c <- IO.inspect(a),
>> d <- {a, b} do
>> ...
>> end
>>
>> That *should* print out "1" immediately because c only has a dependency
>> on a and is unrelated to b, however, your implementation does not. I've
>> been discussing some of these edge cases with José, Luke Imhoff, et al for
>> a few months and slowly built up this gist.
>> <https://gist.github.com/hunterboerner/96bc022d1ac131c21c4ce7f12756d0c3> As
>> you can see, that gist is a complete mess so Luke and I began to rewrite it
>> in https://github.com/kronicdeth/async_with. John Wahba proposed an
>> alternative implementation to me at ElixirConf that eliminates the need to
>> compute a dependency graph so we're going to be implementing that version
>> sometime soon and share it here when we have a working version on GitHub.
>>
>> - Theron
>>
>> On Fri, Sep 8, 2017 at 5:07 AM Fernando Tapia Rico <[email protected]>
>> wrote:
>>
>>> I wanted to share with you guys https://github.com/fertapric/async_with,
>>> which is the implementation of "async with", one of the "José goes crazy"
>>> ideas mentioned in this proposal :) I hope you like it!
>>>
>>> Any feedback or help is more than welcome!
>>>
>>> On Thursday, November 19, 2015 at 4:54:15 PM UTC+1, José Valim wrote:
>>>
>>>> Please open up an issue. If we support such, we would need to support
>>>> it both on "comprehensions" and "with".
>>>>
>>>
>>>>
>>>>
>>>> *José Valim*
>>>> www.plataformatec.com.br
>>>> Skype: jv.ptec
>>>> Founder and Director of R&D
>>>>
>>>> On Thu, Nov 19, 2015 at 5:05 PM, Vincent Siliakus <[email protected]>
>>>> wrote:
>>>>
>>>>> Op zaterdag 14 november 2015 13:57:08 UTC+1 schreef José Valim:
>>>>>>
>>>>>> The implementation is already in master thanks to lexmag. :)
>>>>>>
>>>>>>
>>>>> Works great, I love this addition!
>>>>>
>>>>> I was wondering though since the current implementation uses 'case'
>>>>> internally, if support for something like this is still planned:
>>>>>
>>>>> with x when x < 2 <- 4, do: :ok
>>>>> => 4
>>>>>
>>>>> with x when x > 2 <- 4, do: :ok
>>>>> => :ok
>>>>>
>>>>> I think this would allow even more clunky, nested code to be rewritten
>>>>> much more elegantly with 'with'.
>>>>>
>>>>>
>>>> --
>>> 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/9a5fb079-07a7-44d4-ab3d-d6ab1ab8a9bd%40googlegroups.com
>>>
>>> <https://groups.google.com/d/msgid/elixir-lang-core/9a5fb079-07a7-44d4-ab3d-d6ab1ab8a9bd%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
--
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/6177548b-1e49-4bfb-add1-2a142fae6ac4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
