Skip to content

Handle component type determination more robustly in confTool.py

Teng Jian Khoo requested to merge khoo/athena:Trig-newJO-diff-MC into main

While trying to diff legacy JO vs newJO HLT jsons, we hit a ton of spurious differences due to default value recognition failing mainly on private tools, as the full genealogy for the tool names did not appear in the map of component name to type. This attempts to fix this by recording types both for the bare name and for the one including all ancestors. Possible chance of conflict here, in principle.

A side effect of this is that when CA and legacy differ due to one being default and the other non-default, the latter is suppressed and hence not displayed, but instances of such comparisons are much more rare. It might be possible to retrieve the defaults in such cases, but this is left for future development.

More for debugging purposes, filtered out more expressions from the type/name parsing that look implausible as component specifications.

Tagging @fwinkl, @mark, @tbold as people who might be able to help sanity-check these changes.

Merge request reports