Add a unique ID to Interval properties, allow multiple intervals with the same (start, end)

[pleath]

Add methods to get and remove an interval from a collection by ID. Use ID as a third key to determine order in cases of identical (start, end). "delete(start, end)" will remove all (start, end) intervals from the collection. Each such delete will be serialized as an independent operation.

[pleath]

Replaces #6279.

[msfluid-bot]

■ @fluidframework/base-host: No change

Metric Name	Baseline Size	Compare Size	Size Diff
main.js	178.64 KB	178.64 KB	■ No change
Total Size	178.64 KB	178.64 KB	■ No change

⯅ @fluid-example/bundle-size-tests: +1.83 KB

Metric Name	Baseline Size	Compare Size	Size Diff
container.js	150.96 KB	150.96 KB	■ No change
map.js	48.69 KB	48.69 KB	■ No change
matrix.js	147.17 KB	147.33 KB	⯅ +172 Bytes
odspDriver.js	197.23 KB	197.23 KB	■ No change
odspPrefetchSnapshot.js	28.17 KB	28.17 KB	■ No change
sharedString.js	163.55 KB	165.21 KB	⯅ +1.66 KB
Total Size	735.76 KB	737.59 KB	⯅ +1.83 KB

---

Baseline commit: eec1d66

Generated by 🚫 dangerJS against ac0e1c1

[pleath]

Addresses #6108, #6109.

[anthony-murphy]

    it("propagates", async () => {

can you add some multi client tests

---

Refers to: packages/test/test-end-to-end-tests/src/test/sharedInterval.spec.ts:337 in aa65640. [](commit_id = aa656409b2466d834de50a8fbad920c314eec614, deletion_comment = False)

[anthony-murphy]

a.addPropertySet(b.properties);

how does this, defaultIntervalConflictResolver behave with local vs remote conflict? do both clients end up with the same id, or will different id's never conflict

---

Refers to: packages/dds/sequence/src/intervalCollection.ts:288 in aa65640. [](commit_id = aa656409b2466d834de50a8fbad920c314eec614, deletion_comment = False)

[anthony-murphy]

    it("propagates", async () => {

it would be good to add some scenario focus tests. currently you really only test the local client, as there is only one container, and you never load a new container to test rehydration

---

In reply to: 861874250

---

Refers to: packages/test/test-end-to-end-tests/src/test/sharedInterval.spec.ts:337 in aa65640. [](commit_id = aa656409b2466d834de50a8fbad920c314eec614, deletion_comment = False)

[pleath]

#6407 (comment)

I do see a client receiving the relevant ops in the test I've been adding to. At least the non-local add(serialized) and delete(serialized) getting called to receive the ops. Can you point me to an example of code that loads a new container to test rehydration?

[pleath]

#6407 (comment)

Yeah, different ID's won't conflict. The compare function will compare ID's if (start, end) are the same.

[anthony-murphy]

sequenceNumber

we would usually call this referenceSequenceNumber, as sequenceNumbers are provided by the server #WontFix

[anthony-murphy]

looks to be pre-existing

[pleath]

Do you think the member name in ISerializedInterval should be changed?

[anthony-murphy]

addInternal

should the above add method call this #Closed

[anthony-murphy]

looks like you moved some methods which make the merge gross, but i'm not sure why anything in add needed to change

[anthony-murphy]

add used to call addInternal, which would hit the case below

[pleath]

I changed it to make the local and non-local cases more distinct. I had to do that for delete. It does make debugging easier if addInternal is only called in the case where an op comes over the wire. Sorry about the merge, though.

[anthony-murphy]

it doesn't actually work this way. you need to actually set the reserved id field in props on load from the snapshot, as in the case of shared inverval on a shared string the endpoints can change over time as modifications happen.

overall, i think we might be tackling too many things is this pr, which is making it hard to get any of them in.

[pleath]

Updating to handle the load case, which is the one I wasn't handling.

[anthony-murphy]