Description
Ideally, a regression test should be isolated and depend only on the specific tool it is meant to test.
For example, some tests in CTS call global_placement and/or detailed_placement, which is unnecessary. In fact, DPL (detailed_placement) is currently invoked by the regression tests of several different tools, including cts, grt, odb, psm, ram, rsz, tap, and OpenROAD.
Consequently, PR #10226 (which introduces changes to DPL) requires modifications to over 170 files, mostly updating golden regression targets. This level of maintenance overhead should not be necessary.
Suggested Solution
To address this, we should run the prerequisite commands once to generate an intermediate DEF file. This file can then be stored and used as the direct input for the target tool's regression test, allowing us to safely remove the upstream tool calls.
Additional Context
No response
Description
Ideally, a regression test should be isolated and depend only on the specific tool it is meant to test.
For example, some tests in CTS call global_placement and/or detailed_placement, which is unnecessary. In fact, DPL (detailed_placement) is currently invoked by the regression tests of several different tools, including cts, grt, odb, psm, ram, rsz, tap, and OpenROAD.
Consequently, PR #10226 (which introduces changes to DPL) requires modifications to over 170 files, mostly updating golden regression targets. This level of maintenance overhead should not be necessary.
Suggested Solution
To address this, we should run the prerequisite commands once to generate an intermediate DEF file. This file can then be stored and used as the direct input for the target tool's regression test, allowing us to safely remove the upstream tool calls.
Additional Context
No response