Create ZoneOffsetCalculator

[robertbastian]

#5466

Supersedes #5515

[sffc]

Good work! Unsure on some of the details in datagen but the rest looks great.

[sffc]

Issue: This is the opposite of MetazonePeriodV1, and in your data structs, you end up with a bunch of i32::MAX instead of 0, which will make the two data structs more difficult to merge. Is there a reason you did it this way?

[robertbastian]

TZDB defines its data like this.

[sffc]

If you can do it in 20-30 minutes, I think I prefer storing them the same way as Metazones. If you can't do it in that amount of time, we'll just leave it a the follow-up issue.

[robertbastian]

I tried and it's very messy. The rules are not guaranteed to be ordered, so you have to assemble everything first, and then convert end times to start times. For this conversion you need to keep track of the offset at that point. Not a 20 minute thing.

[sffc]

ok, let's file a follow-up issue to merge the two markers and then we'll have to do this when we merge them.

[sffc]

Observation: MetazonePeriodV1 has

  "aqcas": {
    "0": "auwe",
    "20930040": "case",
    "21128580": "auwe",
    "21995640": "case",
    "22164060": "auwe",
    "24617760": "case",
    "25345020": "auwe",
    "25647600": "case",
    "25879200": "auwe",
    "26168820": "case",
    "26393280": "auwe",
    "26695681": "case",
    "26927340": "auwe",
    "27219841": "case",
    "27451500": "auwe",
    "27744001": "case",
    "27971520": "auwe"
  },

And ZoneOffsetPeriodV1 has

  "aqcas": {
    "20930520": [
      64,
      0
    ],
    "21129240": [
      88,
      0
    ],
    "21996120": [
      64,
      0
    ],
    "22164060": [
      88,
      0
    ],
    "24618240": [
      64,
      0
    ],
    "25345680": [
      88,
      0
    ],
    "25648080": [
      64,
      0
    ],
    "25879860": [
      88,
      0
    ],
    "26169300": [
      64,
      0
    ],
    "26393940": [
      88,
      0
    ],
    "26696161": [
      64,
      0
    ],
    "26928000": [
      88,
      0
    ],
    "27220321": [
      64,
      0
    ],
    "27452160": [
      88,
      0
    ],
    "27744481": [
      64,
      0
    ],
    "27972180": [
      88,
      0
    ],
    "2147483647": [
      64,
      0
    ]
  },

which seem highly correlated. There are some that match exactly, such as 22164060. There are others that are virtually the same, such as 21128580 and 21129240, differing by 11 hours, which seems more likely to be a rounding / calculation / data error than an actual difference in when the two events occurred (metazone switch and offset switch).

[sffc]

@nekevss and @nordzilla may want to review.

[sffc]

Now that you are downloading the TZDB from the GitHub tag, you don't need the tests/data/tzdb directory, right?

[robertbastian]

Now I need it, unless you want cargo test to download it.

[sffc]

oh it's needed for the same reason that we have a subset of CLDR JSON in the repo?

[robertbastian]

yes