Open Bug 1587382 Opened 2 months ago Updated 25 days ago

Provide application as a new perf signature property

Categories

(Tree Management :: Perfherder, task, P3)

Tracking

(Not tracked)

People

(Reporter: igoldan, Unassigned)

References

(Depends on 2 open bugs, Blocks 2 open bugs)

Details

(Keywords: meta)

Currently, performance_signatures have the following properties which define them & together ensure their uniqueness:

  • repository
  • framework
  • platform
  • option_collection
  • suite
  • test
  • last_updated
  • extra_options

We need to add a new browser property, so we can (for example) easily bring the Chrome/Chromium perf data variants, when we're viewing the Firefox ones.

I would suggest the more generic application rather than browser, and also consider an application_version.

See also bug 1587576, which adds some support for obtaining browser metadata.

Priority: -- → P3

(In reply to Dave Hunt [:davehunt] [he/him] ⌚BST from comment #1)

I would suggest the more generic application rather than browser,

Good idea.

and also consider an application_version.

My gut feeling says we cannot support application_version during this quarter. But will do some research to properly confirm that.

Still, if we've reached this subject... It's reasonable to assume that application is a data decoupling factor.
But I consider application_version to not be a data decoupling factor. After we, let's say update Chrome's version used in our CI, we should still see data points collected under the same signature. That way, we'll be able to be alerted on changes like these & label them as test harness changes.
After all, that's how we've handled updates to mitmproxy, when we've switched from 2.0.2 to 4.0.4. Bebe, hopefully I'm right here. Could you please confirm this, if you have a better memory than mine?

Dave, am I right about this assumption?

Flags: needinfo?(dave.hunt)
Flags: needinfo?(fstrugariu)
Summary: Provide browser as a new perf signature property → Provide application as a new perf signature property

(In reply to Ionuț Goldan [:igoldan], Performance Sheriff from comment #3)

Dave, am I right about this assumption?

I agree, let's revisit application_version in the future if desired.

Flags: needinfo?(dave.hunt)
Flags: needinfo?(fstrugariu)
You need to log in before you can comment on or make changes to this bug.