Stream: shorthand
Topic: Not agreeing with an error message
David Pyke (Oct 19 2021 at 14:35):
I would disagree with this statement: Cardinality constraints cannot widen the cardinality. 0..0 is wider than 1..*.
David Pyke (Oct 19 2021 at 14:35):
It's an error, because I'm trying to lock out a required element but it's not wider... 0..* would be...
Chris Moesel (Oct 19 2021 at 14:49):
Yeah, I agree that the message seems a little off. It's "widening" it on the min side (if you think of it as a range on a number line), but narrowing it on the max side. Another way of saying it would be that "Cardinality constraints cannot decrease an existing min or increase an existing max" -- but I wonder if there is a still a more succinct way of saying that. Any ideas?
David Pyke (Oct 19 2021 at 15:13):
I think that would be best, it's not succinct but it explains the problem well.
Elliot Silver (Oct 19 2021 at 16:05):
"Cardinality constraints must match or be tighter than existing constraints."
David Pyke (Oct 19 2021 at 16:05):
0..0 is tighter than 1..*
Elliot Silver (Oct 19 2021 at 16:05):
"must match or be within"
Chris Moesel (Oct 19 2021 at 16:08):
So... "Cardinality constraints must match or be within the existing cardinality. 0..0 is not fully within the cardinality 1..*" ?
David Pyke (Oct 19 2021 at 16:09):
That works!
Elliot Silver (Oct 19 2021 at 16:09):
Yup.
Elliot Silver (Oct 19 2021 at 16:10):
Although now that I read the whole thing, you could shorten it to "Cardinality 0..0 is not fully within the cardinality 1..*"
Elliot Silver (Oct 19 2021 at 16:11):
(And I suppose if I wanted to get nit picky, 1..1 is not fully within 1..* -- proper subsets and all.)
Gino Canessa (Oct 19 2021 at 16:11):
If you're adding a specific case for 0..0 when the base is 1.., you could also go with something along the lines of 'Cannot prohibit required element...'
Chris Moesel (Oct 19 2021 at 16:33):
@Gino Canessa - If we can find one general phrase that works across the board, I'd prefer that over special-casing this message. While using terms like "prohibited" and "required" reads well in English, it would mean that authors would need to know that by "prohibited" we're referring to 0..0 and by "required" we're referring to 1... By just using the cards directly, I think it's easier to see the relation to the actual FSH statement.
@Elliot Silver - I was using "fully within" because it is actually partially within but partially within is not good enough. BUT... I agree that it might be misunderstood by nerdy math types -- so let's drop it.
How about "Cannot constrain cardinality to 0..0, as it does not fit within the original 1..* cardinality"?
Gino Canessa (Oct 19 2021 at 17:01):
Chris: fair =) If there is additional documentation around the error I think adding some of that kind of text may be useful. E.g., that the error is commonly caused by trying to prohibit a required element or require a prohibited one, etc.
Chris Moesel (Oct 21 2021 at 11:25):
For those who wish to track this change, it is SUSHI#938.
Chris Moesel (Nov 12 2021 at 20:50):
We've improved the error message in SUSHI 2.2.0.
Last updated: Apr 12 2022 at 19:14 UTC
