Consecutive similar markers should be aggregated somehow




Developer Tools: Performance Tools (Profiler/Timeline)
2 years ago
2 years ago


(Reporter: Sami Jaktholm, Unassigned)


(Blocks: 1 bug)

Firefox Tracking Flags

(Not tracked)



(2 attachments)



2 years ago
Created attachment 8613206 [details]

1. Open the attached HTML file.
2. Open profiler and start profiling.
3. Let the recording run for 10-15 seconds.
4. Stop the recording.
5. Look for folded 'Function call' markers and click on a long one.

What happens: The sidebar shows that the function call took [length of the folded marker] seconds while really there was multiple calls a second apart and each took ~0.01ms.

See attached the attached screenshot (especially the upper one is quite confusing).

Comment 1

2 years ago
Created attachment 8613207 [details]
Duplicate of this bug: 1170386
These were backed out in bug 1169439 -- need to come up with a good UX story for these.
Summary: The duration for the parent node of folded markers is misleading → Consecutive similar markers should be aggregated somehow
Keeping everything on a single row sounds like a good idea to me. Basically, maintaing the previous implementation, but don't show the "parent" marker as a single contiguous rectangle.
Check out how Chrome does rasterizing markers -- a bit different as these are on a diff thread, but they have a metamarker "Rasterize x7", and these markers don't have any actionable data. As more meta is added to style/reflow markers, I'd imagine I'd want them uncollapsed by default, but we'll see
On IRC, wilsonpage pinged me of confusion with collapsing of JS markers, since we use different names for generic and specific:

Function Call>
-  Script Tag
-  Script Tag

I think we should reserve any kind of similar marker collapsing to only a few types (recalc, layout, paint)
See Also: → bug 1233831
Triaging. Filter on ADRENOCORTICOTROPIC (yes).
Priority: -- → P3
You need to log in before you can comment on or make changes to this bug.