MergeTree: Make SortedSegmentSet Iterative and Add Some Tests

[anthony-murphy]

Description

This is related to #11603 and #1009. Like the merge tree walk functions the recursiveness here was also causing perf issue when trying during undo, which frequently adds and removes items from a SortedSegmentSet via a tracking group. There were no directly test for SortedSegmentSet, as it was mainly tested via tracking groups, so adding a couple tests. Additionally, using a SortedSegmentSet with an object that provided a segments {segment: ISegment} was broken previously. This change also fixes that issue by allowing multiple elements with a match ordinal.

AB#151

[anthony-murphy]

i was having problems with my editor defaulting to tabs, this seemed to help. editor.detectIndentation will still override it, but it must use insertSpaces when it is not sure, like in a new file

[msfluid-bot]

⯅ @fluid-example/bundle-size-tests: +16.53 KB

Metric Name	Baseline Size	Compare Size	Size Diff
aqueduct.js	392.45 KB	394.42 KB	⯅ +1.98 KB
connectionState.js	680 Bytes	680 Bytes	■ No change
containerRuntime.js	191.92 KB	197.48 KB	⯅ +5.55 KB
loader.js	151.12 KB	151.06 KB	⯆ -57 Bytes
map.js	42.63 KB	47.38 KB	⯅ +4.75 KB
matrix.js	131.63 KB	134.64 KB	⯅ +3.01 KB
odspDriver.js	150.23 KB	150.11 KB	⯆ -127 Bytes
odspPrefetchSnapshot.js	38.39 KB	38.35 KB	⯆ -41 Bytes
sharedString.js	152.42 KB	153.88 KB	⯅ +1.46 KB
Total Size	1.25 MB	1.27 MB	⯅ +16.53 KB

---

Baseline commit: fe10534

Generated by 🚫 dangerJS against 7c80d42

[Abe27342]

looks good. I wonder if we could get some better code re-use between SortedSegmentSet and partial lengths. The fact that segments' ordinals can change but they must remain stable is still enough for them to satisfy reasonable comparator invariants, and from there it's a pretty generic data structure.

[anthony-murphy]

looks good. I wonder if we could get some better code re-use between SortedSegmentSet and partial lengths. The fact that segments' ordinals can change but they must remain stable is still enough for them to satisfy reasonable comparator invariants, and from there it's a pretty generic data structure.

interesting! i hadn't thought about that. you should open an issue and describe the overlap between the two, as if we can share code i'd love to get it scheduled

[Abe27342]

looks good. I wonder if we could get some better code re-use between SortedSegmentSet and partial lengths. The fact that segments' ordinals can change but they must remain stable is still enough for them to satisfy reasonable comparator invariants, and from there it's a pretty generic data structure.

interesting! i hadn't thought about that. you should open an issue and describe the overlap between the two, as if we can share code i'd love to get it scheduled

sounds good. I tracked this internally in AB#1663 and put it in next sprint. Could be a nice ramp-up task and works toward our code size quality goals.