Vendor Glean SDK v64.5.1
Categories
(Data Platform and Tools :: Glean: SDK, task, P1)
Tracking
(firefox142 fixed)
| Tracking | Status | |
|---|---|---|
| firefox142 | --- | fixed |
People
(Reporter: chutten|PTO, Assigned: chutten|PTO)
References
(Blocks 1 open bug, Regressed 1 open bug)
Details
Attachments
(6 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 |
It has a new dual_labeled_counter metric type that we'll be wanting to use presently, plus a fix to label counting: Release Notes.
| Assignee | ||
Comment 1•6 months ago
|
||
| Assignee | ||
Comment 2•6 months ago
|
||
| Assignee | ||
Comment 3•6 months ago
|
||
No profiler markers because no impl MetricType for
glean::DualLabeledCounterMetric (bug 1973287)
| Assignee | ||
Comment 4•6 months ago
|
||
Also includes JOG support.
Submetrics assume they're always submetrics of labeled_* metrics,
but now counters could also be submetrics of a dual_labeled_counter.
Take a page from FOG Rust's book and encode the parent metric's type as part of
the submetrics' type.
Open q'ns:
- Should we call it CounterMetricImpl<C> and
using CounterMetric = CounterMetricImpl<CounterType::eBaseOrLabeled>;? - Should the FFI distinguish between the add/test methods on the submetric
and the get method on the parent? They're all dual_labeled_counter_*.
| Assignee | ||
Comment 5•6 months ago
|
||
Supports mirroring to histograms with keyed: true and
kind: {categorical|boolean}.
| Assignee | ||
Comment 6•6 months ago
|
||
Backed out for causing run failures and build bustages.
- Backout link
- Push with failures
- Push with failures build bustages
- Failure Log run failures
- Failure Log build bustages
| Assignee | ||
Comment 10•6 months ago
|
||
Current status:
So it turns out that when you build the whole shebang, gradle gets the latest glean_parser and latest Glean SDK and builds successfully because the latest changes to Glean to support dual_labeled_counter aren't breaking outside of the boundary scribed by those two tools together. The interface between those two tools did change, though, and that causes problems when a task uses some prebuilt components and some non-prebuilt components together, like when you try just bookmarks sync or the PGO run.
The solution we're pursuing is the same as if this were a capital-B Breaking change: vendor the Glean+glean_parser update manually into application-services and coordinate the inclusion into mozilla-central of that nightly build of application-services to happen at the same time as we vendor Glean+glean_parser into m-c. (An alternate solution would be to somehow explain to Gradle that it can't mix and match like this, but we're going to stick with the tried-and-true method this time.)
(( The team has resolved to treat changes between the Glean SDK and glean_parser as Breaking for the purposes of deciding when to make a major semver increment from now on to hopefully prevent this from catching us off guard again. ))
When that's all landed, likely tomorrow, we can lay the FOG dual_labeled_counter implementation patches overtop and be good to land.
Updated•6 months ago
|
Updated•6 months ago
|
Comment 11•6 months ago
|
||
Comment 12•6 months ago
|
||
| bugherder | ||
https://hg.mozilla.org/mozilla-central/rev/33bad266b7d4
https://hg.mozilla.org/mozilla-central/rev/fbce9d85a2de
Comment 13•6 months ago
|
||
Oops, forgot to add the keep-open tag. There's some patches we now need to land as a followup. That's back in :chutten's hands.
Comment 14•6 months ago
|
||
Comment 15•5 months ago
|
||
| bugherder | ||
https://hg.mozilla.org/mozilla-central/rev/71ae6fc503a7
https://hg.mozilla.org/mozilla-central/rev/6c4c4dd329d1
https://hg.mozilla.org/mozilla-central/rev/4f6194b053e6
https://hg.mozilla.org/mozilla-central/rev/5d6916df1ab3
Description
•