Provide DateTime.from_iso8601!/2

[Mário Guimarães]

Hello,

all "from_x" functions in DateTime support their throw exception version, but the function "from_iso8601/2" which returns an error tuple.

Doesn't it make sense, because of convenience and completeness, to provide a bang version of this function?

See https://hexdocs.pm/elixir/DateTime.html#from_iso8601/2

Thanks

Mário

[Mário Guimarães]

The proposed function could be implemented like this

  def from_iso8601!(string, calendar \\ Calendar.ISO) do
    case DateTime.from_iso8601(string, calendar) do
      {:ok, datetime, offset} ->
        {datetime, offset}

      {:error, reason} ->
        raise ArgumentError,
              "cannot convert #{string} to datetime, reason: #{inspect(reason)}"
    end
  end

[Andrea Leopardi]

I can't see a reason why not to add this. I think a PR would be welcome!

Andrea Leopardi

an.le...@gmail.com

--
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/1b2ea98e-24b0-4829-8943-82ca3add955c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Fernando Tapia Rico]

This has been suggested in the past. For example https://github.com/elixir-lang/elixir/pull/6514

The main reason why there is not a bang variant is because the `offset` information would be dropped.

On Wednesday, June 19, 2019 at 10:13:39 AM UTC+2, Andrea Leopardi wrote:
> I can't see a reason why not to add this. I think a PR would be welcome!
>
>
>
>
>
> Andrea Leopardi
> an.le...@gmail.com
>
>
>
> On Mon, Jun 17, 2019 at 4:31 PM Mário Guimarães <mario.lui...@gmail.com> wrote:
>
> The proposed function could be implemented like this
>
>
>
>
>   def from_iso8601!(string, calendar \\ Calendar.ISO) do
>     case DateTime.from_iso8601(string, calendar) do
>       {:ok, datetime, offset} ->
>         {datetime, offset}
>
>       {:error, reason} ->
>         raise ArgumentError,
>               "cannot convert #{string} to datetime, reason: #{inspect(reason)}"
>     end
>   end
>
>
>
>
> segunda-feira, 17 de Junho de 2019 às 15:18:29 UTC+1, Mário Guimarães escreveu:
> Hello,
>
>
> all "from_x" functions in DateTime support their throw exception version, but the function "from_iso8601/2" which returns an error tuple.
>
>
> Doesn't it make sense, because of convenience and completeness, to provide a bang version of this function?
>
>
> See https://hexdocs.pm/elixir/DateTime.html#from_iso8601/2
>
>
> Thanks
> Mário
>
>
>
>
> --
>
> 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-l...@googlegroups.com.

[Andrea Leopardi]

We don't drop the offset information in the suggested implementation. And pattern matching on {datetime, offset} instead of {:ok, datetime, offset} still has the benefit that it has to match otherwise a nice error is returned (which doesn't happen if you match on {:ok, datetime, offset} and get a MatchError).

Andrea Leopardi

an.le...@gmail.com

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/f8d56771-d618-4ece-be85-4f4f740fe928%40googlegroups.com.

[Fernando Tapia Rico]

Oh! Sorry, I replied too soon, the example contains the offset.

I’m not sure if users would expect just the datetime in the bang version :(

[Andrea Leopardi]

Well, it's documented, and if the non-bang variant returns the offset I think users would expect that?

Tbh to me the benefits (better error, mirroring all other functions) outweigh the downsides (returning a tuple instead of a datetime like other functions).

Andrea Leopardi

an.le...@gmail.com

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/dab80e48-9121-41dd-8d19-dc9658096af1%40googlegroups.com.

[Chris]

This seems like a nice cleanup of a minor blemish.

To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/CAM9Rf%2BJDNy3PztCikgbXnNs_pcecRBKamN-%2Bm8xoPZ_CoFEYwQ%40mail.gmail.com.