chore: replace deprecated object_store Path::child with join

object_store 0.13 deprecates Path::child in favor of .join()/.clone().join()
(exact behavioral equivalent: single percent-encoded segment; join takes self
by value). Fixes 8 call sites that would fail CI under -D warnings. Also adds
UPGRADE-NOTES capturing the to_timestamp upstream-kept decision and the
encrypted_parquet feature-flag finding.

Author: pauldix

UPGRADE-NOTES-df53.1-arrow58.3.md

@@ -0,0 +1,121 @@
+# Upgrade notes — DataFusion 50.1→53.1 / arrow 56.2→58.3
+
+InfluxData-internal notes for the `upgrade-df-ver5310-a` branch. Records decisions and
+behavioral deltas that are **not** obvious from the code or commit messages. Keep this file
+updated for any further changes; **delete it before any PR to apache** (influx-internal scratch).
+
+Versions: DataFusion **53.1.0**, arrow/parquet **58.3.0** (skip-pinned ahead of 53.1's declared
+58.0, by request), sqlparser 0.61, object_store 0.13.x, MSRV 1.88, edition 2024.
+
+---
+
+## 1. `to_timestamp` execution-timezone behavior — UPSTREAM KEPT (revert deliberately skipped)
+
+**Decision (2026-05-27, Paul Dix): keep upstream 53.1 behavior. Do NOT re-apply the revert.**
+Documented here because it may matter later.
+
+### What this is
+The prior 52.5 patch set carried influx commit `bb43b6890`,
+*Revert "Respect execution timezone in to_timestamp and related functions (#19078)"*.
+Upstream #19078 (landed 2026-01-05, in 53.1.0) changed the `to_timestamp*()` family so that
+**naïve (timezone-free) timestamp strings are interpreted in the configured
+`datafusion.execution.time_zone`** (UTC only when that option is `None`), and results carry the
+execution timezone. Before #19078, naïve strings were always interpreted as **UTC**.
+
+The 52.5 revert restored the always-UTC behavior. During the 53.1 upgrade the revert was **skipped**
+(it conflicts in ~14 regions against 53.1's reworked `to_timestamp.rs` / `common.rs` / `mod.rs` /
+`to_date.rs` / `to_unixtime.rs` / `macros.rs`, and 53.1 also added Float16/Float32/Decimal numeric
+input support via #19663 that must not be lost).
+
+### Current state on this branch
+- Runs **upstream 53.1 behavior**: naïve `to_timestamp('...')` is interpreted in
+  `datafusion.execution.time_zone`.
+- `datafusion/sqllogictest/test_files/to_timestamp_timezone.slt` is present and **passing**.
+- The upstream implementation bug in #19078 (#20223, scalar-float broadcast) was fixed by #20224
+  on 2026-02-12, which is in 53.1 — so there is no outstanding correctness bug to work around.
+
+### Why this is safe to keep (and the risk if assumptions change)
+This is a no-op for InfluxDB **as long as IOx does not set a non-UTC
+`datafusion.execution.time_zone`**. If any future IOx code path sets a non-UTC execution timezone,
+this upstream behavior will **silently shift parsed naïve timestamps** away from UTC — a correctness
+change for InfluxDB queries. **If that ever happens, re-derive the revert against 53.1** (force
+naïve input → UTC while preserving 53.1's numeric-type support); do not replay the 52.5-era diff.
+Touch points if re-deriving: `to_timestamp.rs`, `common.rs`, `mod.rs`, `to_date.rs`,
+`to_unixtime.rs`, `macros.rs`; reconcile `to_timestamp_timezone.slt` and
+`datetime/timestamps.slt`.
+
+---
+
+## 2. `encrypted_parquet.slt` — NOT an arrow 58.3 regression (false alarm in the original handoff)
+
+**Resolution (2026-05-27): no code change, no test edit, no version pin-back needed.**
+
+The original handoff flagged this test as failing due to a "genuine upstream change between parquet
+58.0 and 58.3." That diagnosis was **incorrect**. Root cause: the test was run **without** the
+`parquet_encryption` feature.
+
+### What actually happens
+Parquet encryption (read *and* write) is gated behind `#[cfg(feature = "parquet_encryption")]`.
+With the feature **off**, the `format.crypto.*` table options are silently ignored, so the
+"encrypted" files are written as **plaintext** (footer magic `PAR1`, plaintext schema/column names
+on disk). The final step of the test then reads those plaintext files without keys and **succeeds**,
+so the `query error … decryption properties were not provided` assertion fails with
+"expected to fail, but actually succeed."
+
+This is independent of the arrow/parquet version — the feature gate behaves identically under 58.0
+and 58.3. (Verified by reasoning about the `cfg` gate; the on-disk files confirm plaintext output
+when the feature is off.)
+
+### How to run it correctly
+```bash
+cargo test -p datafusion-sqllogictest --test sqllogictests --features parquet_encryption -- encrypted_parquet
+```
+With the feature **on**: files are written with an encrypted footer (magic `PARE`, no plaintext
+column names on disk), the keyless read errors as expected, and the test **passes**. Verified on
+this branch under arrow/parquet 58.3.
+
+**Operational note:** the bare `cargo test -p datafusion-sqllogictest --test sqllogictests` command
+fails on `encrypted_parquet.slt` on *any* build without the feature (including pristine 53.1 base).
+Always pass `--features parquet_encryption` when running the slt suite, or run the full
+`cargo test --workspace` (where feature unification across workspace members enables it).
+
+---
+
+## 3. `nth_value` / `first_value` per-kind signatures — DO NOT UNDO
+
+The "improve OneOf signature diagnostics" commit backports apache #21032 (not in 53.1). It gives
+`nth_value`/`first_value` **per-kind** signatures (`Signature::any(1)` / `Signature::any(2)`) rather
+than 53.1's `one_of([Nullary, Any(1), Any(2)])`. Reverting to `one_of` re-breaks 5 `errors.slt`
+cases (e.g. `one_of` wrongly lets `first_value(c5, 2)` succeed). **Keep per-kind.**
+
+---
+
+## 4. Pre-existing upstream test (not our regression)
+
+`datafusion-expr` → `expr_rewriter::order_by::test::rewrite_sort_cols_by_agg_alias` is flagged in the
+handoff as failing on the pristine 53.1.0 tag (feature/env-dependent), unrelated to our patches or
+the arrow bump. It did **not** surface in the full `cargo test --workspace` run on this branch
+(2026-05-27, exit 0, 0 failures across all suites). Leave it; flag upstream if it recurs.
+
+---
+
+## 5. Verification status (2026-05-27)
+
+- `cargo test --workspace` — **PASS** (exit 0, 0 failures across all suites).
+- `encrypted_parquet.slt` with `--features parquet_encryption` — **PASS**.
+- `cargo clippy --workspace --all-targets -- -D warnings` — **PASS** (exit 0) after the fix below.
+- **Fixed:** 8 `object_store::path::Path::child` deprecation warnings (object_store 0.13 bump)
+  that would fail CI's `-D warnings`. Replaced with the maintainer-sanctioned `.join()` /
+  `.clone().join()` (exact behavioral equivalent — single percent-encoded segment; `join` takes
+  `self` by value so a `.clone()` is added where the receiver was borrowed). Sites:
+  `datasource/src/write/demux.rs` (3), `datasource/src/url.rs` (1 line, 2 calls),
+  `core/tests/sql/path_partition.rs` (1), `core/src/datasource/physical_plan/parquet.rs` (1).
+  Affected tests re-run and green (datasource lib, `test_prefix_path`, `path_partition`,
+  `parquet_exec_with_error`).
+
+## 6. Finishing checklist
+- [x] §1 to_timestamp revert — resolved: keep upstream, documented above.
+- [x] §2 encrypted_parquet — resolved: false alarm, run with `--features parquet_encryption`.
+- [x] Full `cargo test --workspace` — green.
+- [x] `cargo clippy --workspace --all-targets -- -D warnings` clean (fixed 8 `Path::child` deprecations).
+- [ ] Delete `HANDOFF-df-upgrade-ver5310.md` and this `UPGRADE-NOTES-*.md` before any PR to apache.

datafusion/core/src/datasource/physical_plan/parquet.rs

@@ -1703,7 +1703,7 @@ mod tests {
         let state = session_ctx.state();
         let location = Path::from_filesystem_path(".")
             .unwrap()
-            .child("invalid.parquet");
+            .join("invalid.parquet");
 
         let partitioned_file = PartitionedFile::new_from_meta(ObjectMeta {
             location,

datafusion/core/tests/sql/path_partition.rs

@@ -766,7 +766,7 @@ impl ObjectStore for MirroringObjectStore {
             };
 
             if parts.next().is_some() {
-                common_prefixes.insert(prefix.child(common_prefix));
+                common_prefixes.insert(prefix.clone().join(common_prefix));
             } else {
                 let object = ObjectMeta {
                     location: k.clone(),

datafusion/datasource/src/url.rs

@@ -536,7 +536,7 @@ mod tests {
         let root = root.to_string_lossy();
 
         let url = ListingTableUrl::parse(root).unwrap();
-        let child = url.prefix.child("partition").child("file");
+        let child = url.prefix.clone().join("partition").join("file");
 
         let prefix: Vec<_> = url.strip_prefix(&child).unwrap().collect();
         assert_eq!(prefix, vec!["partition", "file"]);

datafusion/datasource/src/write/demux.rs

@@ -260,7 +260,8 @@ fn generate_file_path(
     if !single_file_output {
         base_output_path
             .prefix()
-            .child(format!("{write_id}_{part_idx}.{file_extension}"))
+            .clone()
+            .join(format!("{write_id}_{part_idx}.{file_extension}"))
     } else {
         base_output_path.prefix().to_owned()
     }
@@ -588,8 +589,8 @@ fn compute_hive_style_file_path(
 ) -> Path {
     let mut file_path = base_output_path.prefix().clone();
     for j in 0..part_key.len() {
-        file_path = file_path.child(format!("{}={}", partition_by[j].0, part_key[j]));
+        file_path = file_path.join(format!("{}={}", partition_by[j].0, part_key[j]));
     }
 
-    file_path.child(format!("{write_id}.{file_extension}"))
+    file_path.join(format!("{write_id}.{file_extension}"))
 }