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
I run into a situation that I wanted to test local schemas changes in setuptools and see if they work first before cutting new releases to validate-pyproject.
So my strategy was to have local .spec.json files and load them with --tool ...=file://....
However I run into a problem:
[ERROR] SchemaWithDuplicatedId: All schemas used in the validator MUST define a unique toplevel `"$id"`
This happens because the plugin is loaded from an entry-point and from tool.
So I thought about to alternatives not mutually exclusive:
Add an option to the CLI to disable all plugins loaded from entry-points1
Footnotes
This might also be useful for improving reproducibility. But it becomes more complicated to handle if we also allow "extra" (non-tool, "side-car") plugins to be loaded from entry-points, in the sense that we would have a bunch of open questions... ↩
The text was updated successfully, but these errors were encountered:
I run into a situation that I wanted to test local schemas changes in setuptools and see if they work first before cutting new releases to
validate-pyproject
.So my strategy was to have local
.spec.json
files and load them with--tool ...=file://...
.However I run into a problem:
[ERROR] SchemaWithDuplicatedId: All schemas used in the validator MUST define a unique toplevel `"$id"`
This happens because the plugin is loaded from an entry-point and from
tool
.So I thought about to alternatives not mutually exclusive:
--tool
to always overwrite whatever plugin is loaded from entry-points (I believe this is done in Allow overwrite when schemas refer to the sametool
#175).Footnotes
This might also be useful for improving reproducibility. But it becomes more complicated to handle if we also allow "extra" (non-tool, "side-car") plugins to be loaded from entry-points, in the sense that we would have a bunch of open questions... ↩
The text was updated successfully, but these errors were encountered: