Avoid generator in _ViewInstrumentMatch.collect() #3035
Conversation
lancetarn
changed the title
lancetarn
changed the title
|
This could be a bug fix, @lancetarn, please add a CHANGELOG entry. ✌️ |
CHANGELOG updated! Thanks @ocelotl You (or someone) might need to kick off CI again for it. |
lancetarn
force-pushed
the
3021-metrics-iterator-to-sequence
branch
from
0a930e8 to
10db92b
Compare
|
It looks like one of the tornado contrib tests assumes I've reproduced this by checkout out the contrib repo and pointing CORE_REPO at my forked version; if there is a better way I'm interested but that seems ok. |
|
Okay I have set open-telemetry/opentelemetry-python-contrib#1456 up to solve the problem. Hopefully that was a reasonable thing to do. |
|
You will also have to update the Contrib SHA here with last commit hash that fixed the issue on contrib repo https://github.com/open-telemetry/opentelemetry-python/blob/main/.github/workflows/test.yml#L13 |
Would this later just have to be reverted? I was intending to just let the fix land over there first, then re-run these. |
|
@lancetarn |
lancetarn
force-pushed
the
3021-metrics-iterator-to-sequence
branch
from
10db92b to
1e49add
Compare
- Fix open-telemetry#3021 - Remove metric branch reference from CONTRIBUTING.md
lancetarn
force-pushed
the
3021-metrics-iterator-to-sequence
branch
from
1e49add to
2be1687
Compare
|
@lzchen Now I see, thank you. I'll point it back at |
|
@lancetarn |
…/github.com/shalevr/opentelemetry-python into decode-values-from-otel-resource-attributes * 'decode-values-from-otel-resource-attributes' of https://github.com/shalevr/opentelemetry-python: Avoid generator in _ViewInstrumentMatch.collect() (open-telemetry#3035)
Description
Alter
_ViewInstrumentMatch.collect()to return a realizedListofdata_pointsrather than aGenerator. This probably has some memory implications, but I don't have a good handle on how impactful they are likely to be.Also updates CONTRIBUTING.md to remove mention of
metricsbranch.Fixes #3021
Type of change
This does change a return type, but it appears effectively internal and nothing within the SDK seemed to attempt to use the result in an incompatible way. The tests were actually assuming it was an
Iteratorrather than just anIterable, I believe, so updating them felt like a fix.How Has This Been Tested?
Installed in editable mode with 3.9 and an external script to send signals to the console and a remote collector. Not a load test, but everything was collected and sent as expected.
Does This PR Require a Contrib Repo Change?
I did remove reference to the
metricsbranch in CONTRIBUTING.md, but that does not seem relevant.Checklist: