Closed paulemmanuel-garcia closed 2 years ago
Have you tried computing the median by first aggregating for the target name? This should allow you to accurately compute the build times for each of your targets: https://github.com/spotify/XCMetrics/blob/ac8ce19e52ed95cab6f4579ba3e07bdafd55698d/proto/xcmetrics/v1/xcmetrics.proto#L34
Hello,
Thank you very much for your answer.
In the build_targets
table I have lines for (non exhaustive list for example):
name |
---|
Write all-product-headers.yaml |
Compile Swift source files |
Process product packaging |
Run custom shell script 'Run Script' |
I don't see any "proper" target name. I think that is why I discarded this table when doing my queries because it did not seem relevant.
Is it something expected?
I think you're looking at the build steps instead of build targets: https://github.com/spotify/XCMetrics/blob/ac8ce19e52ed95cab6f4579ba3e07bdafd55698d/proto/xcmetrics/v1/xcmetrics.proto#L50-L67
You can see a bunch of example data in the tables in the introductory video: https://www.youtube.com/watch?v=kgdeqcYJKfo&t=2s
Thank you for your answer, I was able to find the proper target name in the build_targets
but I still have all those intermediary steps in this table.
Thank you very much I think it can be closed.
Hello,
When querying XCMetrics for the main target app build time I have also the build time for the RichNotificationExtension. When doing median calculation it makes the results flaky as the extension takes a lot less time to build.
Is there a way to differentiate those two? Do you have suggestion on how to workaround that issue?
On our CI we are using a cloud based provider and it seems that the machine name is changing for each pipeline, so by aggregating first by machine name I get coherent results. But this solution won't work for local developer machines. Furthermore, as we are monitoring success builds for the same build I can have the main app succeeding but the extension failing and it will not be a failed build.
Thank you very much