Add alpha feature to output big data passing file path instead of task run name #993
Conversation
Tomcli
changed the title
| appended_taskrun_path_step = _get_base_step('output-taskrun-path') | ||
| if len(appended_taskrun_path_step['command']) <= 2: | ||
| appended_taskrun_path_step['command'].append('') | ||
| appended_taskrun_path_step['command'][-1] += 'echo -n "%s/%s/%s" > $(results.%s-path.path)\n' % \ |
There was a problem hiding this comment.
I was wondering --- could a prefix (or suffix, or format in general) for paths and values be generalized somehow? My immediate thought would be user-generated outputs and complains about not being able to use FOO (standardized into foo) and FOO_PATH (standardized into foo-path) in the same task.
There was a problem hiding this comment.
I parametrize as a python global variable and renamed the suffix to -datap (short for data path). If you have a better naming idea I can update it in the global variable.
| param.get('name').endswith("-trname") or | ||
| param.get('name').endswith("-path") |
There was a problem hiding this comment.
Using suffixes to do some magic in the compiler seems tricky. If the user creates a task with a result named e.g. asset-path, it will be interpreted as a path to asset result (which does or does not exist), is that correct?
There was a problem hiding this comment.
These functions are to remove any generated results that are pushed into the pipelineloop or generated by the custom taskspec. I think in the newer Tekton 0.37, we can specify a result type other than string. So we can also looking into using result type to specify these special results.
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: Tomcli, yhwang The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Which issue is resolved by this Pull Request:
Resolves #987
Description of your changes:
Currently Big data passing output the task name for incoming tasks to construct the result path as its input. We can to change the output itself to a path so that the incoming tasks won't have extra logics to construct paths.
Things we found that each task has its own workspace prefix, so the workspace variable is still coded as part of the consuming path.
To enable this feature, run
Environment tested:
python --version):tkn version):kubectl version):/etc/os-release):Checklist: