pierre-warnier commented on issue #9183:
URL: https://github.com/apache/arrow-rs/issues/9183#issuecomment-6973091354
Following up on @alamb's first step: here is where `chrono` appears in the
public API today.
Method: `cargo public-api -sss --all-features` (v0.52.0, nightly 2026-11-17)
on every workspace crate that depends on `chrono`, at 5d46410 (60.2.0), then
grep for `chrono`. `arrow-arith`, `arrow-csv`, `arrow-json`, `parquet`,
`parquet-variant-compute` and `parquet-variant-json` have no `chrono` type in
their signatures. `arrow` only re-exports the items below.
| Crate | Item | chrono type |
|---|---|---|
| arrow-array | `temporal_conversions::*` (25 fns: `as_date`, `as_datetime`,
`as_datetime_with_timezone`, `as_time`, `as_duration`, `date32_to_datetime`,
`timestamp_*_to_datetime`, `time*_to_time`, `time_to_time*`,
`duration_*_to_duration`, ...) | `NaiveDate`, `NaiveDateTime`, `NaiveTime`,
`Duration`, `DateTime<Tz>` |
| arrow-array |
`PrimitiveArray::value_as_{date,datetime,datetime_with_tz,time,duration}` |
`NaiveDate`, `NaiveDateTime`, `DateTime<Tz>`, `NaiveTime`, `Duration` |
| arrow-array | `ArrowTimestampType::{make_value, from_naive_datetime,
from_datetime<Tz: TimeZone>}` (trait methods, implemented by the 4
`Timestamp*Type`s) | `NaiveDateTime`, `DateTime<Tz>`, bound `chrono::TimeZone` |
| arrow-array | `Date32Type::{to_naive_date, to_naive_date_opt,
from_naive_date}`, `Date64Type::{to_naive_date_opt, from_naive_date}` |
`NaiveDate` |
| arrow-array | `impl chrono::TimeZone for timezone::Tz` (`offset_from_*`),
`impl chrono::Offset for timezone::TzOffset` (`fix`) | `TimeZone`, `Offset`,
`FixedOffset`, `LocalResult`, `NaiveDate(Time)` |
| arrow-array | `timezone::GapInfo` (`new`, pub fields `begin`, `end`) |
`NaiveDateTime`, `DateTime<Tz>` |
| arrow-array, arrow | cargo feature `chrono-tz` | feature name |
| arrow-cast | `parse::string_to_datetime<T: TimeZone>` | `DateTime<T>`,
bound `chrono::TimeZone` |
| parquet-variant | `Variant::{Date, Time, TimestampMicros, TimestampNanos,
TimestampNtzMicros, TimestampNtzNanos}` (enum payloads) | `NaiveDate`,
`NaiveTime`, `DateTime<Utc>`, `NaiveDateTime` |
| parquet-variant | `Variant::as_{naive_date, time_utc, timestamp_micros,
timestamp_nanos, timestamp_ntz_micros, timestamp_ntz_nanos}`, `From<NaiveDate /
NaiveTime / NaiveDateTime / DateTime<Utc>> for Variant` | same |
| parquet_derive | generated code for `NaiveDate` / `NaiveDateTime` fields
calls `::chrono::...` | `NaiveDate`, `NaiveDateTime` |
Not types, but tied to chrono's behaviour: the format strings in
`arrow_cast::display::FormatOptions::with_{date,datetime,timestamp,timestamp_tz,time}_format`
and in `arrow_csv::WriterBuilder::with_*_format` are parsed as chrono
`strftime` items. Moving the internals to `jiff::fmt::strtime` would change
what some of those strings mean, so they need their own compatibility note.
Most of these are free functions or inherent methods, so a `chrono` feature
with `#[deprecated]` items is straightforward. The harder ones are the trait
impls on `Tz`/`TzOffset`, the `ArrowTimestampType` trait methods, and the
`Variant` enum payloads, where a feature flag would change the shape of a type
rather than add items.
If it helps, I can open the feature-gating PRs, one per crate starting with
`arrow-array`, with jiff equivalents alongside the deprecated chrono items.
I'll wait for a go on the shape (feature name, default on or off) before
starting.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]