STI defines several types of test subjects: .rpm, .repo, .qcow, ocl, docker..
It also mentions that there could be several artifacts defined via a "concatenated string". There is uncertainty in how it should actually work.
1) Concatenated string - what is the delimiter? Comma? space?
2) How to handle the group of subjects? Run test for each of them separately? Group by type?
Consider the case TEST_SUBJECTS=bash.rpm,python3.rpm,local.repo. Our current implementation of inventory scripts will create one test environment with local.repo and two rpm packages installed.
TEST_SUBJECTS=bash.rpm,python3.rpm,local.repo
But it won't work for TEST_SUBJECTS=test.qcow2,some_other_test.qcow2. Do we need to create two test environments? Or fail with error?
TEST_SUBJECTS=test.qcow2,some_other_test.qcow2
Maybe implement two delimiters: ; for groups and , inside a group?
;
,
For example:
TEST_SUBJECTS=bash.rpm,python3.rpm,local.repo;test.qcow2
Metadata Update from @bookwar: - Issue tagged with: STI, discussion
After some thinking about it, I propose to simplify it for now.
Let's support:
1) Comma-separated list of .rpm and .repo files
It will be treated as one test environment, as it is now.
2) One .qcow2 or .qcow2c image 3) One .oci image 4) One docker: image
There are a lot of ways this can be extended, but i suggest that we don't do it yet until there is an actual use-case.
Login to comment on this ticket.