Rich wires #523
Rich wires #523
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## develop #523 +/- ##
===========================================
- Coverage 90.20% 90.20% -0.01%
===========================================
Files 102 102
Lines 6321 6310 -11
===========================================
- Hits 5702 5692 -10
+ Misses 619 618 -1
Continue to review full report in Codecov by Sentry.
|
| classical_out or set(), | ||
| classical_in or set(), | ||
| ): | ||
| if any( |
There was a problem hiding this comment.
I don't think we need this check since the type hints already make it explicit that's what we expect
There was a problem hiding this comment.
I think we need it because otherwise one can pass an unsorted tuple, or list. Like (3,2,1) and then the modes are screwed up. To guard against that we would have to cast to a set, but I'd rather not do that by default because then implicit stuff happens.
There was a problem hiding this comment.
But we type hint it as a set does that mean we should update the type hint?
There was a problem hiding this comment.
What change do you have in mind? Here the hint says set[int] | None and the method raises if it isn't that.
There was a problem hiding this comment.
Right the type hint says its set | None so we shouldn't be expecting something like an unsorted tuple or list to be passed in. If we want to support passing in an unsorted tuple or list then the type hint should be updated to reflect that. If not, then the check shouldn't be necessary as the type hint is already implying it is set | None.
There was a problem hiding this comment.
As I'm sure you know, the type hint does not enforce anything. Raising an exception in this case guards against running into bugs because of an ordered data structure is passed in when out of order. If this is your only concern about this PR can you accept this check and approve it?
Wires have been challenging to maintain due to evolving requirements.
This PR introduces a better abstraction for Wire that stores a collection of
QuantumWireandClassicalWiredataclasses. Metadata can be added to these dataclasses without modifying the Wires object’s handling of indices and dictionaries. Many methods, like__matmul__, are greatly simplified.