Just when I thought I was ready to complete the back-merge and soft freeze, @aclum and I discovered that an NMDC contributor had entered data into MongoDB where slot S was mandatory and required format F1.
I believe that the contributor legitimately didn't have data to fill S, so they added vales in format F2. In some cases F2 can be coerced to F1 and appears to be valid. In other cases, F1 is not only invalid, but it causes tools that are part of the overall validation suite to crash.
I think all of us try clever workarounds in the construction of the schema or construction of data records sometimes. I certainly have on a few occasions.
We need cultural and technical barriers to clever workarounds. That could include
better documentation
stricter slot definitions by default (instead of adding strictness after the fact)
requesting LinkML tools to better handle edge cases
But most of all, if you are facing a schema challenge and you know you are using a workaround, contact me, @sierra-moxon, @pkalita-lbl or @cmungall as soon as possible.
Just when I thought I was ready to complete the back-merge and soft freeze, @aclum and I discovered that an NMDC contributor had entered data into MongoDB where slot
S
was mandatory and required formatF1
.I believe that the contributor legitimately didn't have data to fill
S
, so they added vales in formatF2
. In some casesF2
can be coerced toF1
and appears to be valid. In other cases,F1
is not only invalid, but it causes tools that are part of the overall validation suite to crash.I think all of us try clever workarounds in the construction of the schema or construction of data records sometimes. I certainly have on a few occasions.
We need cultural and technical barriers to clever workarounds. That could include
But most of all, if you are facing a schema challenge and you know you are using a workaround, contact me, @sierra-moxon, @pkalita-lbl or @cmungall as soon as possible.