DO NOT MERGE: trial new schema w/ input#998
DO NOT MERGE: trial new schema w/ input#998petertseng wants to merge 1 commit intoexercism:masterfrom petertseng:schema-with-input
Conversation
what we wanted to see |
|
fun fact: the one other exercise that is valid under the new schema, with no additional changes? ETL. |
| , "property" : { "$ref": "#/definitions/property" } | ||
| , "input" : { "$ref": "#/definitions/input" } | ||
| , "expected" : { "$ref": "#/definitions/expected" } | ||
| } |
There was a problem hiding this comment.
Should this disallow additional properties?
There was a problem hiding this comment.
I currently can't think of a need for other properties. So, great question and/or idea, let's do it.
|
To make this transition, it is:
I wonder if I should follow this plan: For each exercise: Its canonical-data.json file will be verified using the new schema, unless the directory contains a file named |
|
Do we need the extra file? |
|
This would fail miserably when someone misspells the `"input"` key.
|
|
I currently count five exercises compliant. |
|
currently exercises/acronym/canonical-data.json valid |
|
@petertseng Could you re-run the tool to see which exercises are now compliant? Is it also possible to see which exercise are not compliant? |
OK, rebased on master. Since I won't be around to report when it finishes, I'll have to leave it to you to read the output when it finishes.
Yes. |
|
Thanks @petertseng! The current status is as follows. Valid:
Invalid:
|
If we do this:
Since I predict I as a reviewer may make a mistake in reading the CI output carefully, I preferred to make the intention explicit instead of implicit. Since to me it makes a clearer history if we avoid having commits that look like "convert to new schema" (PR gets merged) "oops, actually convert to new schema correctly this time", I am biased toward making the review as error-free as possible, despite that we burden implementors with having to remember to However, since to date I have neither been an implementor nor a reviewer of PRs converting to the new schema, I think I had better defer my decision to those who have been in such positions. |
|
What should we do with test cases where there is no input? An example of this is the |
|
Among the choices I thought of were:
I think I would prefer
|
👍 |
Agreed! |
This schema was proposed and accepted in #996
No description provided.