Open gfyrag opened 1 year ago
hey there i haven't played with it myself yet but my understanding from the integration coverage docs and the latest set of go test
flags is they you simply set a GOCOVERDIR
environment variable to have your precompiled binaries emit coverage information. each invocation of the binary will generate a new file that you can inspect and merge with go tool covdata
and i would assume that the go community has done the work to ensure that concurrent invocations of the binary do not generate issue so you can simply run GOCOVERDIR=/path/to/dir ginkgo -p
and things should work. My only recommendation is that you use an absolute path for GOCOVERDIR
as Ginkgo will change the working directory when invoking specs in subdirectories.
Give it a try!
I have a Similar Issue.
Here Running go test works but i have used Labels to exclude few tests which i am unable to do.
export GOCOVERDIR=`pwd`
ginkgo -r '--label-filter=!setup'
This is not Generating Binary Coverage Files which can be processed by covdata.
Eg go tool covdata percent -i=.
Come back on this issue (I had forget it to be honnest...).
I have tried your suggestion @onsi , and the directory stay empty. I used that command:
GOCOVERDIR=$PWD/coverage/e2e ginkgo -tags it -cover -p ./test/...
Hi!
I'm using Ginkgo for integration testings and would like to use the flag
-test.gocoverdir
introduced with Go 1.20. If using plain "go test", it works as expected, but not with the ginkgo cli. Unfortunately, using "go test" involve no parallel testing. I tried to "go test" with the flags -ginkgo.parallel.XXX, but ginkgo complains about it. Not surprising as i'm not supposed to do that.Do you have any plan tu support it? Or maybe a workaround?