Associating the helper process with Port instances doesn't make sense, since there should only ever be a single helper process, and that process should only be managed by the parent process of run-webkit-tests.
<rdar://problem/78575542>
Created attachment 429900 [details] Patch
Comment on attachment 429900 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=429900&action=review > Tools/Scripts/webkitpy/port/base.py:963 > it isn't, or it isn't available. If a port overrides start_helper() > it must override this routine as well.""" We may want to update the comment here because MacPort does not overwrite this routine. > Tools/Scripts/webkitpy/port/base.py:964 > + if Port.helper: Do we want this to be 'self.helper', otherwise, MacPort.stop_helper() will do nothing.
Created attachment 429953 [details] Patch
Created attachment 429954 [details] Patch
Created attachment 430280 [details] Patch
Comment on attachment 430280 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=430280&action=review r=me with comment. > Tools/ChangeLog:12 > + (Port.stop_helper): All classes should share the same code to stop the helper process. Might worth mentioning we are doing this because popen object is not pickleable.
Created attachment 430300 [details] Patch
Committed r278332 (238365@main): <https://commits.webkit.org/238365@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 430300 [details].