Closed ErnestFistozz closed 2 years ago
Hi @ErnestFistozz.
I think that, in this case, this is correct.
If you notice the project's recent builds:
The last four (4) appear to have been triggered by a once-monthly cron
job. I would venture to say that, while this triggered a new build in CI, there were no changes to the repo (on master
) and therefore, each new build was created using the same commit sha.
This tracks with the fifth (5th) most recent build (Build: 6338), the fifth (5th) in your results, having a different commit sha and different information from Github (commit
, committer
, type
, via
, etc).
As for the created_at
field, that pertains to the build, which is a different build in CI, so no surprise there.
But you have a good point about the covered_percent
being different, which doesn't make a lot of sense for what should be identical builds.
That said, there are a number of reasons why test coverage results could be different for the identical codebase with no changes to code or tests. In this case, one might be time-sensitive tests.
Another thing I noticed that could account for unexpected coverage changes is that, while Builds 6338-6972 each have two (parallel) jobs:
Example: https://coveralls.io/builds/47730656
Build 7257 only has one: https://coveralls.io/builds/50412252
Which could indicate a failed build, or something else, like a change to CI config.
Hi @afinetooth
Thanks, I don't know why I didn't think of cron
jobs. but you are absolutely correct it is a cron job.
Quick question though, when it comes to the number of reasons
link, this points me to a tool called CodeCov
. (which is another tool i am using). Does this mean CodeCov
and Coveralls
are the same from a design perspective.
Hi @ErnestFistozz. Yes, Codecov is a competitor of Coveralls in the test-coverage-as-a-service space. We've been around a bit longer and focus on open-source as part of our original mission. They have great documentation and cover that topic well.
Completed!
Hi,
I am trying to collect coverage from one of the repository in the Apache Foundation. The coverage data is specifically for the
master
branch. The issue Branch Coverage Through API describes how to retrieve the desired response. My endpoint is:https://coveralls.io/github/apache/rocketmq.json?page=1&branch=master
Below is the response i receive. However, the response is not correct because the
created_at
ANDcovered_percent
properties changes YET thecommit_sha
property remains the same. That is definitely aBUG
because agit commit hash
cannot be the same for different commits. Each hash is unique.I am particularly referring to
first
element to thefourth
element in the builds array i.ebuilds[0], builds[1], builds[2]
andbuilds[3]
.NB : I have redacted the author's email and name