[RFC] Add a mechanism to set the chosen case for SwitchProducer #35510
Conversation
|
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-35510/25705
|
|
A new Pull Request was created by @makortel (Matti Kortelainen) for master. It involves the following packages:
@makortel, @smuzaffar, @cmsbuild, @Dr15Jones can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
|
@cmsbuild, please test |
|
|
||
| The decision can be "forced" with 'foo.setCase_("case1")' and | ||
| unset with 'foo.setCase_(None)'. This setting persists through | ||
| dumpPython() and picking. |
There was a problem hiding this comment.
| dumpPython() and picking. | |
| dumpPython() and pickling. |
| c = cms.int32(2) | ||
| ) | ||
| ) | ||
| ).setCase_('test1') |
There was a problem hiding this comment.
Dumping the "set case" like this looks a bit weird (setCase_() returns self so it should work), but doing it along
foo = SwitchProducerTest(
...
)
foo.setCase_(...)would require knowing the module label (which is not available here), or extending the functionality of specialImportRegistry (would need to think more what exactly that would imply).
I also considered
SwitchProducerTest(
test1 = ...,
test2 = ....,
setCase_ = cms.untracked.string('test1')
)but that appeared to require adding special cases for setCase_ in many places complicating the implementation even more. Also given that this setting should(?) be somewhat special case, and that it is not passed on to C++, making it a function call felt better choice than a "real parameter".
|
test parameters:
|
|
@cmsbuild, please test |
|
+1 Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-a9008a/19328/summary.html Comparison SummarySummary:
|
|
Closing in favor of #36699 |
PR description:
This PR attempts to resolve to add a mechanism to "force a chosen case" for a SwitchProducer following the idea in #31760 (comment).
Opening now as a RFC, because I'm not fully satisfied with the naming I came up, and I'm not fully convinced of the approach (not that I would have really better ideas).
PR validation:
Framework unit tests run.