[Proposal] Non-naive DateTime sigil

[Gregory McIntyre]

Love Elixir! So good!

Recently we found ourselves writing a lot of DateTime literals to feed into our unit tests like this:

DateTime.from_naive!(~N[2016-05-05 01:01:42], "Etc/UTC")

Is there a simpler way to specify human readable timestamp literals in code? Could there be? I'm not sure what the best sigil would be, but something like this?

~Z[2016-05-05T01:01:42Z]

(That is ISO 8601 format.)

Love and peace and many thanks,

Greg

[Wojtek Mach]

I think that'd be a nice addition. Even if it does not eventually land in Elixir, anyone can define the `~Z` (or whatever) sigil just for tests as shown at the end of the post.

I usually deal with just `NaiveDateTime`s but on the rare occasion that I used `DateTime` not having the sigil wasn't as bad as the incredibly verbose inspect output of DateTime's (the whole struct). Thus, if there was an unambiguous text representation (that contains timezone) which could be used in both the sigil and inspect, I think that would be a great addition to the language. While inspect could handle any struct (assuming it was correctly constructed, leaving this to the developer), sigil probably would only work for Etc/UTC, until Elixir ships with tzdata.

defmodule FooCase do
  use ExUnit.CaseTemplate

  using do
    quote do
      def sigil_Z(binary, _opts) do
        {:ok, datetime, 0} = DateTime.from_iso8601(binary)
        datetime
      end
    end
  end
end


defmodule FooTest do
  use FooCase


  test "the truth" do
    assert ~Z[2015-01-23T23:50:07Z] == DateTime.from_naive!(~N[2015-01-23T23:50:07], "Etc/UTC")
  end
end

[Michał Muskała]

I'm not 100% convinced about the sigil since it would only work with UTC.

What I would love to see, however, is an Inspect implementation for DateTime. The structs are extremely hard to read and see what's happening. Even having a simple inspect implementation should make it much easier.

Michał.

--
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/d59e56aa-dde9-426b-9303-a62f2576e396%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Wojtek Mach]

There's DateTime.to_string/1 which seems perfect to use, I'm looking into it:

iex(1)> DateTime.utc_now
#DateTime<2017-06-27 14:46:54.331343Z>

iex(3)> %DateTime{year: 2000, month: 2, day: 29, zone_abbr: "CET", hour: 23, minute: 0, second: 7, microsecond: {0, 0}, utc_offset: 3600, std_offset: 0, time_zone: "Europe/Warsaw"}
#DateTime<2000-02-29 23:00:07+01:00 CET Europe/Warsaw>

[Wojtek Mach]

For anyone interested, I've implemented the ~Z sigil to experiment with in a personal project: https://gist.github.com/wojtekmach/b66d1bd1eadf17cc57f7ae64f0f6d4b1

This allows me to do:

$ iex -S mix
iex(1)> DateTime.utc_now
~Z[2017-09-08 21:04:17.404239]
iex(2)> Calendar.DateTime.shift_zone!(~Z[2017-09-08 21:04:17.404239], "America/New_York")
#DateTime<2017-09-08 17:04:17.404239-04:00 EDT America/New_York>

The Code.compiler_options.ignore_module_conflict is obviously a hack, so looking forward to hearing if it's a big no-no :)