Support labeled_{custom|memory|timing}_distribution metric types in FOG + GIFFT
Categories
(Toolkit :: Telemetry, enhancement, P2)
Tracking
()
Tracking | Status | |
---|---|---|
firefox130 | --- | fixed |
People
(Reporter: chutten, Assigned: chutten)
References
Details
Attachments
(8 files, 3 obsolete files)
48 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review |
After bug 1657947 adds support for labeled_{custom|memory|timing}_distribution
metrics, and after that code gets into a release, we'll need to have that release vendored into m-c and perform the necessary work to support the new metric types in FOG:
- Rust, C++, and JS APIs (and tests) - includes IPC
- GIFFT support (and tests) - these metric types provide APIs which'll allow us to mirror to Legacy keyed linear/exponential histograms (which is why bug 1885380 gets a See Also link)
- JOG support (and tests) - though the initial consumers are expected to be in platform, we expect frontend uses to abound as well
Assignee | ||
Comment 2•1 year ago
|
||
Assignee | ||
Comment 3•1 year ago
|
||
Assignee | ||
Comment 4•1 year ago
|
||
This makes me sad, since these are for quantities that only permit unsigned
values. But this is what the Glean SDK was forced to use by early Kotlin,
and it simplifies codegen a little now that the signed types are re-exposed
by LabeledMetricData... so I guess we suffer.
Assignee | ||
Comment 5•1 year ago
|
||
Assignee | ||
Comment 6•1 year ago
|
||
Assignee | ||
Comment 7•1 year ago
|
||
Assignee | ||
Comment 8•1 year ago
|
||
This makes me sad, since these are for quantities that only permit unsigned
values. But this is what the Glean SDK was forced to use by early Kotlin,
and it simplifies codegen a little now that the signed types are re-exposed
by LabeledMetricData... so I guess we suffer.
Assignee | ||
Comment 9•1 year ago
|
||
Assignee | ||
Comment 10•1 year ago
|
||
Assignee | ||
Comment 11•1 year ago
|
||
Assignee | ||
Comment 12•1 year ago
|
||
This one is a little more complex because we need to support GIFFT for pure
Rust consumers (bug 1751448).
To do that we need to split the GIFFT impl from the core impl of FOG's Rust
timing_distribution implementation. That makes it so that the labeled_*-
specific submetric impl can dispatch to the core impl and call to GIFFT
independently.
It also makes the GIFFT code in timing_distribution.rs a little nicer
(to my eyes), which is a nice plus.
Updated•1 year ago
|
Updated•1 year ago
|
Updated•1 year ago
|
Comment 13•1 year ago
|
||
Comment 14•1 year ago
|
||
Comment 15•1 year ago
|
||
bugherder |
https://hg.mozilla.org/mozilla-central/rev/cc66fc79f49e
https://hg.mozilla.org/mozilla-central/rev/cd2bf02d8b29
https://hg.mozilla.org/mozilla-central/rev/63852846489b
https://hg.mozilla.org/mozilla-central/rev/99c22c70694e
https://hg.mozilla.org/mozilla-central/rev/88858e864eb7
https://hg.mozilla.org/mozilla-central/rev/c8925e80d2b9
https://hg.mozilla.org/mozilla-central/rev/820858301aa8
https://hg.mozilla.org/mozilla-central/rev/cfa2a54bceae
https://hg.mozilla.org/mozilla-central/rev/9c8bb4b86267
Description
•