Agentless Does Not Run the Full Test Suite as Described in the Paper/README

[WuYff]

Description

According to the paper (Section 3.3.2), Agentless is expected to run all existing tests in the repository to identify passing tests before selecting regression tests. Specifically, the paper states:

"AGENTLESS first runs all the existing tests in the repository to identify a set of passing tests that successfully pass in the original codebase … AGENTLESS provides the list of passing tests to the LLM…"

Similarly, the README states:

"We select tests from the complete list of tests which can pass in the original repository."

However, upon reviewing the code, it appears that Agentless directly calls the get_test_directives function to extract specific test files from the developer's test patch rather than running the entire test suite.

Request

Could you provide more details/implementation about how to run the full test suite in the repository as described in the paper/README?