Page MenuHomePhabricator

Goldilocks 2.0: Benchmark single test
Open, p3: NextPublic7 Energy Points (d+f*r)


What is the exact feature or behavior required?

In Goldilocks 1, you can run comparative benchmarks between two tests, but can't run a benchmark with a single test entered.

It would be nice to be able to benchmark a single test.

One or more example use cases for the feature, including sample expected input and output.

Example: working on optimizing a function, a user would like to run a "before" and "after" test for the specific logic they are working on, to see if they made an improvement.

Side note: might also be handy to be able to save benchmark results to some kind of file, too.

An explanation of why the feature is needed. This is important to prioritization. “It would be nice if...” takes a backseat to “This common scenario doesn’t work without...”.

It would be handy, because otherwise "before" and "after" style comparison would require duplicating the code called and having both states present in the codebase at once.

Any initial ideas you have on implementation (if applicable).


Task Type
Proposed Urgency
g3: Major
d4: Within Sprint
f3: Off-Road
r1: Trivial
Volatility (Caught At)
Not a Bug
Not a Bug/Unknown

Related Objects


Event Timeline

jcmcdonald edited projects, added Goldilocks [Project]; removed Unknown Object (Project).Jun 19 2021, 10:14 AM
jcmcdonald added a parent task: Restricted Maniphest Task.Jul 3 2021, 5:28 PM
jcmcdonald changed the task status from Proposed to Open.Jul 3 2021, 6:34 PM
jcmcdonald triaged this task as p3: Next priority.
jcmcdonald removed a project: Announce [Control].
jcmcdonald changed Gravity from Triage Gravity to g3: Major.
jcmcdonald set Distance to d4: Within Sprint.
jcmcdonald changed Friction from Triage Friction to f3: Off-Road.
jcmcdonald changed Relativity from Triage Relativity to r1: Trivial.
jcmcdonald set the point value for this task to 7.
jcmcdonald added a subscriber: jcmcdonald.