sql: Add support for reverse loose index scans

[rohany]

Implements support for reverse scans by the IndexSkipTableReader. Supports interleaved tables, but scans on these tables will be slower than in the forward direction.

Depends on #38442

Resolves #38416

[cockroach-teamcity]

This change is 

[solongordon]

Good stuff! Comments could use a bit of clarification but otherwise .

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @rohany and @solongordon)

---

pkg/sql/distsqlrun/index_skip_table_reader.go, line 210 at r10 (raw file):

		if !t.spans[t.currentSpan].Valid() {
			t.currentSpan++
			continue

Huh, was this continue a bug?

---

pkg/sql/distsqlrun/index_skip_table_reader.go, line 211 at r10 (raw file):

			t.spans[t.currentSpan].Key = key
		} else {
			// In the case of reverse, this is much easier. The reverse batcher

Did you mean fetcher?

---

pkg/sql/distsqlrun/index_skip_table_reader.go, line 212 at r10 (raw file):

		} else {
			// In the case of reverse, this is much easier. The reverse batcher
			// gives us the key that we got. Since ranges are exclusive, we just

"gives us the key that we got" 🤔

Also maybe better: "Since EndKey is exclusive..."

[rohany]

Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale) (waiting on @solongordon)

---

pkg/sql/distsqlrun/index_skip_table_reader.go, line 210 at r10 (raw file):

Previously, solongordon (Solon) wrote…

Huh, was this continue a bug?

yeah, it was. If the next span was invalid, we wouldn't return the row we were supposed to even though the current span was valid.

---

pkg/sql/distsqlrun/index_skip_table_reader.go, line 211 at r10 (raw file):

Previously, solongordon (Solon) wrote…

Did you mean fetcher?

Done.

---

pkg/sql/distsqlrun/index_skip_table_reader.go, line 212 at r10 (raw file):

Previously, solongordon (Solon) wrote…

"gives us the key that we got" 🤔

Also maybe better: "Since EndKey is exclusive..."

Done.

[rohany]

bors r+

[craig]

Build succeeded

• GitHub CI (Cockroach)