Hey Geoprocessors!
Something interesting I just ran into that I thought I would share. I don't know how many tools this extends to, or how many of you may already know this, or who of you might know other ways around it, but I'm sharing nonetheless.
I have a simple GP tool, published as a web tool to Enterprise 10.9.1 that takes a user-defined point, generates a buffer, and uses SummarizeWithin() to calculate field stats from a hosted feature layer. The first version I built used a hard-coded value table for the sum_fields parameter, i.e.
[['abovepoverty','mean']]
This worked fine when consuming in a web app. The second version of the tool actually took the multi-value choices from another tool parameter, split them, built the value table, and assigned it to a variable, which was then used as the argument to sum_fields, i.e.
stats_to_calc = [[field, 'sum'] for field in summary_fields.split(';')]
SummarizeWithin(
...,
sum_fields=stats_to_calc
)
After publishing this second version, the tool always failed, with
ERROR 100014: Summarize Within failed. Failed to execute (SummarizeWithin).
According to an Esri support doc, this has to do with one of:
- Feature Class name length
- Filepath length
- Field names incorrect/too long
None of these made sense to me. Finally, I decided to try one more code alteration, so now the calculation of the value table for the sum_field parameter happened inside the SummarizeWithin() function call, i.e.
SummarizeWithin(
...,
sum_fields=[[field, 'sum'] for field in summary_fields.split(';')]
)
When I published the tool with this alteration, the tool successfully generates my output buffer and summary table. I am not sure why it only likes this methodology, but that's what I've found.
Anyone insight is of course appreciated 🙂
- Jack C
