Clean up pass over all superfluous hashing happening on the query path #8207
Conversation
teh-cmc
added
⛃ re_datastore
|
Web viewer built successfully. If applicable, you should also test it:
Note: This comment is updated whenever you push a commit. |
|
@rerun-bot full-check |
|
Started a full build: https://github.com/rerun-io/rerun/actions/runs/11974807708 |
teh-cmc
changed the title
|
@rerun-bot full-check |
|
Started a full build: https://github.com/rerun-io/rerun/actions/runs/11975363476 |
|
|
||
| timelines.sort_by(|(field1, _times1), (field2, _times2)| field1.name.cmp(&field2.name)); |
There was a problem hiding this comment.
a bit surprised we care about the order in the transport chunk?
There was a problem hiding this comment.
We do! RecordBatches respect the order of columns so keeping it stable is pretty important, otherwise the UX for end users is pretty awful.
| "37.0 MiB in total", | ||
| "37 B per row", |
There was a problem hiding this comment.
.. hurray...?
So not sorting in some places saved memory? Guess something randomly compacted better?
| } | ||
|
|
||
| /// Retrieve all [`ComponentName`]s in the store. | ||
| pub fn all_components_sorted(&self) -> ComponentNameSet { |
There was a problem hiding this comment.
Maybe we should rename this type to OrderedComponentNameSet to pair it up with UnorderedComponentNameSet, a bit confusing in the context of this PR :/
(or even better imho ComponentNameOrderedSet & ComponentNameUnorderedSet for friendlier autocomplete)
There was a problem hiding this comment.
I wanted to, but the UI code already exposes a SortedComponentSet with slightly different rules... 🤪
|
I want this and none of the questions/comment here are blocking enough to not just merge this. |
#8207) 2x-3x perf improvements in scenes with many entities 😬 Before:  After: 
2x-3x perf improvements in scenes with many entities 😬
Before:
After:
