You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Add support for set and frozenset (and dict?) as native Click types #3036
changed the title [-]Add support for `set` and `frozenset` as native Click types[/-][+]Add support for `set` and `frozenset` (and `dict`?) as native Click types[/+]on Aug 23, 2025
I am unclear what this issue is for. @kdeldycke can you please be more specific. Is it for adding the param types for set, frozenset, dict? I saw the test that shows there is a way to instantiate the types.
I am unclear what this issue is for. @kdeldycke can you please be more specific. Is it for adding the param types for set, frozenset, dict? I saw the test that shows there is a way to instantiate the types.
My goal is to start sparkling these native types in the various unit tests. And just search for edge-cases and inconsistent behavior. A bit like fuzzy testing. If everything's fine then we will have additional tests. If not I will start a discussion and propose code to fix stuff. That's it ! :)
Could you describe what you expect to happen with these types? tuple is pretty clear, it indicates a compound type, and data from the CLI is fixed, not a mutable list.
Could you describe what you expect to happen with these types? tuple is pretty clear, it indicates a compound type, and data from the CLI is fixed, not a mutable list.
I created this ticket to not hide under the rug the concerns raised about these types in #2610, so I could close the later. And also based on some gut feelings. But if this is too speculative for now, we can close it as too vague. I may come back in the future with a more material proposition in the form of a PR.
All the current behavior of Click around these types has been locked down and covered by unittest in #3808 to prevent regression and document the current state of things.
This issue is still open as it target proper conservation and handling of these types, which has not been addressed and will probably involve some breaking changes. As such I move this issue to the major 9.0.0 release.
Activity
[-]Add support for `set` and `frozenset` as native Click types[/-][+]Add support for `set` and `frozenset` (and `dict`?) as native Click types[/+]Rowlando13 commentedon Sep 26, 2025
I am unclear what this issue is for. @kdeldycke can you please be more specific. Is it for adding the param types for set, frozenset, dict? I saw the test that shows there is a way to instantiate the types.
kdeldycke commentedon Sep 29, 2025
My goal is to start sparkling these native types in the various unit tests. And just search for edge-cases and inconsistent behavior. A bit like fuzzy testing. If everything's fine then we will have additional tests. If not I will start a discussion and propose code to fix stuff. That's it ! :)
Rowlando13 commentedon Sep 30, 2025
Got it.
davidism commentedon Sep 30, 2025
Could you describe what you expect to happen with these types?
tupleis pretty clear, it indicates a compound type, and data from the CLI is fixed, not a mutable list.kdeldycke commentedon Oct 8, 2025
I created this ticket to not hide under the rug the concerns raised about these types in #2610, so I could close the later. And also based on some gut feelings. But if this is too speculative for now, we can close it as too vague. I may come back in the future with a more material proposition in the form of a PR.
Test frozen types; Documents type conversion
Test frozen types; Documents type conversion
Test frozen types; Documents type conversion
kdeldycke commentedon Sep 8, 2026
All the current behavior of Click around these types has been locked down and covered by unittest in #3808 to prevent regression and document the current state of things.
This issue is still open as it target proper conservation and handling of these types, which has not been addressed and will probably involve some breaking changes. As such I move this issue to the major 9.0.0 release.