Why does Select By Location (Select Subset) still analyze every feature?

[AlfredBaldenweck]

Version 3.1

Full disclosure, these polygons live in an enterprise GDB

I'm doing a project which involves the following:

• Select a group of polygons that intersect a line
• From these polygons, select a subset that have their center in a certain zone
• From this subset, select a second subset those whose boundary don't touch the line

What I'm finding is that step 3, rather than testing the 40 or so polygons, is testing all 1.4 MILLION. 

Okay, so let's try to problem solve a little bit. What if I make a layer from selection after step 2? It still tests all 1,400,000. 

So I'm a little confused here, because

1) I'm trying to get a subset of a selection, so it stands to reason it would only analyze the subset and

2) It's giving me the warning about how many records I have selected before I run it, so like, why bother telling me that if it's going to be a lie?

For my workflow, it actually was faster to EXPORT after step 2 and then run step 3 on the copy. Like, 33 total seconds vs 3 minutes. I kid you not.

So, cool, and I'm glad I have a solution, but I'm still left with a big ???

What's going on here?

[Robert_LeClair]

@AlfredBaldenweck - question for you - what geoprocessing tools are you using for your workflows?  In many of the GP tools, there is a slider where you can "Use the Selected records" option to only use those selected features in the GP tool.  Buffer, for example, has this option.

[AlfredBaldenweck]

In this case that I'm complaining about, it is literally the three steps I described: Select By Location (1), Select By Location (2), Select By Location (3).

To your point, I think @Bud has complained in the past about how some but not all tools will take the selection into account, but it isn't documented on their help p....

In this case, the tool (Select By Location) not only shows "You have 48 records selected" on the page, but it also is explicitly supposed to be selecting a subset, so I am flummoxed as to why it would check against the whole dataset anyway on Step 3. (As for how I know: 1) It takes 3 minutes and 2) when I made a layer from selection to see if that sped it up, there was a brief graphical error where it showed every other polygon in the dataset after running the select again.

[Robert_LeClair]

Hmmm....so this is interesting.  There was "BUG-000166109 - Select Layer by Attribute tool when run on a large in-memory dataset and returns incorrect results" thta I'm wondering may be the cause of your conundrum.  You've mentioned you're on ArcGIS Pro 3.1.  When the support analyst tested this on ArcGIS Pro 3.2 and 3.3, they were not able to reproduce the workflow.  Question for you - are you able to upgrade your ArcGIS Pro to 3.2 or 3.3?  If not, can you install it within a test environment to see if it's a 3.1 issue or something else?  Please advise. 

[AlfredBaldenweck]

I'm unable to upgrade, unfortunately. It's all managed by the organization.

In any case, the end result is correct, but it seems to check the whole data set, than compare to the subset, rather than just starting with the subset.