If a result (baseline|target|actual) value is present, then it must be a number

[samuele-mattiuzzo]

IATI-Rulesets/rulesets/standard.json

Line 139 in b88c627

{"if": "result/indicator/baseline/@value", "then": "number(result/indicator/baseline/@value) = number(result/indicator/baseline/@value)"},

IATI-Rulesets/rulesets/standard.json

Line 140 in b88c627

{"if": "result/indicator/period/target/@value", "then": "number(result/indicator/period/target/@value) = number(result/indicator/period/target/@value)"},

IATI-Rulesets/rulesets/standard.json

Line 141 in b88c627

{"if": "result/indicator/period/actual/@value", "then": "number(result/indicator/period/actual/@value) = number(result/indicator/period/actual/@value)"}

We're checking for numeric values to be numbers in Ruleset but this should instead be enforced at Schema level. We should remove the rules from the Ruleset and the tests for them.

We should also update the Schema type in the relevant places (example -> xsd:string to xsd:decimal)

[andylolz]

@samuele-mattiuzzo It seems like you might be working through this gist, and creating issues. Is that the case?

[samuele-mattiuzzo]

we have made a copy of the gist and re-assessed some of the rules you pointed out yes.

I'm also logging issues without necessarily:

• having finalised all the discussions around them with the team
• having tagged all the correct versions they need fixing on

[andylolz]

Great! I didn’t hear anything back regarding the gist, so I wasn’t sure whether you’d looked at it. Great to hear you are going through the points 👍 It was quite a lot of work, so I’m pleased it’s useful.

[samuele-mattiuzzo]

Yeah sorry about the silence, but like I said, hectic times :D I forked your gist and used it as a base for our investigations, we also needed to decide whether the effort and time used to fix them was worth it or we should have instead backtracked to pre-January

We learned a bit more about Ruleset, Standard and how to read them through those January changes, so we kept them and decided to fix them. Some are straight forward, some still require some back and forth discussions (eg, this one, as it changes a Schema type effectively enforcing the must be omitted part as a side effect, sort of, so it's sort of sensitive).

Your gist is invaluable btw!

[samuele-mattiuzzo]

How this will be changed (@amy-silcock to 👍 / 👎 )

• we cannot change the Schema because it would prevent 2.02 data from passing 2.03 validation; the number type requiredness is also a should and not a must, one more reason to retain xsd:string as the base type

• the checks in if then will have to be removed completely for this rule (not the tests as they check for the more generic if-then approach)

• this is a 2.03 only change

[samuele-mattiuzzo]

Relevant PR #204

[amy-silcock]

Details of the issue and how we propose to fix it can be found here: https://discuss.iatistandard.org/t/bug-fix-the-new-2-03-result-baseline-and-period-value-rules-are-not-backwards-compatible/1731