Check context more often while sorting.

[ashwin95r]

This change is 

[manishrjain]

Many comments. Address them and then submit.

---

Reviewed 1 of 1 files at r1.
Review status: all files reviewed at latest revision, 3 unresolved discussions.

---

worker/sort.go, line 140 at r1 (raw file):

			// Copy, otherwise it'd affect the destUids and hence the srcUids of Next level.
			tempList := &protos.List{ts.UidMatrix[i].Uids}
			if err := sortByValue(ctx, ts.Attr, ts.Langs, tempList, sType, ts.Desc); err != nil {

This function is accessing so many ts values. Why not just give it the ts struct directly?

---

worker/sort.go, line 232 at r1 (raw file):

			// Intersect every UID list with the index bucket, and update their
			// results (in out).
			err := intersectBucket(ctx, ts, ts.Attr, token, out)

Again, what's the point of giving both ts and then ts.Attr?

If this is simple refactoring, just make it part of this PR.

---

worker/sort.go, line 282 at r1 (raw file):

	go func() {
		// Wait for 3ms before starting
		time.Sleep(3 * time.Millisecond)

Check context right after time.Sleep.

---

Comments from Reviewable

[ashwin95r]

Review status: 0 of 1 files reviewed at latest revision, 3 unresolved discussions.

---

worker/sort.go, line 140 at r1 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

This function is accessing so many ts values. Why not just give it the ts struct directly?

Done.

---

worker/sort.go, line 232 at r1 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Again, what's the point of giving both ts and then ts.Attr?

If this is simple refactoring, just make it part of this PR.

Done.

---

worker/sort.go, line 282 at r1 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Check context right after time.Sleep.

Done.

---

Comments from Reviewable

[manishrjain]

Review status: 0 of 1 files reviewed at latest revision, 1 unresolved discussion.

---

worker/sort.go, line 284 at r2 (raw file):

		time.Sleep(3 * time.Millisecond)
		select {
		case <-ctx.Done():

select case between time.After(3*time.Millisecond) and ctx.Done. Once you're through it, only then do the sortWithoutIndex (it shouldn't be in default).

---

Comments from Reviewable

[manishrjain]

Review status: 0 of 1 files reviewed at latest revision, 3 unresolved discussions.

---

worker/sort.go, line 324 at r2 (raw file):

// intersectBucket intersects every UID list in the UID matrix with the
// indexed bucket.
func intersectBucket(ctx context.Context, ts *protos.SortMessage, token string, out []intersectedList) error {

100 chars.

---

worker/sort.go, line 335 at r2 (raw file):

	key := x.IndexKey(attr, token)
	// Don't put the Index keys in memory.
	pl, decr := posting.Get(key, 1)

How come the group is always 1?

Anyhow, looks like Get doesn't need group. Can you remove that as well?

---

Comments from Reviewable

[manishrjain]

Reviewed 1 of 1 files at r2.
Review status: 0 of 1 files reviewed at latest revision, 3 unresolved discussions.

---

Comments from Reviewable

[ashwin95r]

Review status: 0 of 2 files reviewed at latest revision, 3 unresolved discussions.

---

worker/sort.go, line 284 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

select case between time.After(3*time.Millisecond) and ctx.Done. Once you're through it, only then do the sortWithoutIndex (it shouldn't be in default).

Right! Done.

---

worker/sort.go, line 324 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

100 chars.

Done.

---

worker/sort.go, line 335 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

How come the group is always 1?

Anyhow, looks like Get doesn't need group. Can you remove that as well?

Done.

---

Comments from Reviewable

[manishrjain]

Very important comment. Address before submitting.

---

Reviewed 2 of 2 files at r3.
Review status: all files reviewed at latest revision, 1 unresolved discussion.

---

worker/sort.go, line 285 at r3 (raw file):

			// Wait between ctx chan and time chan.
		case <-ctx.Done():
			resCh <- result{err: ctx.Err()}

No return??? Focus, man!

---

Comments from Reviewable

[ashwin95r]

Review status: 1 of 2 files reviewed at latest revision, 1 unresolved discussion.

---

worker/sort.go, line 285 at r3 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

No return??? Focus, man!

Ah my bad. Done.

---

Comments from Reviewable

[ashwin95r]

Merged.