This is somewhat related to Breaking up long Keywords, though not quite the same. In that ticket, we found the main issue was that we were putting the "Detailed Location" part in where the Subregion 3 information should go. To solve this, we put "NONE" in the Subregion 3 so that the Detailed Location part is in the proper place in the hierarchy.
However, in doing a recent ingest, we were still getting an ingest warning:
<?xml version="1.0" encoding="UTF-8"?><result><concept-id>C1216394131-NSIDC_TS1</concept-id><revision-id>18</revision-id><warnings>After translating item to UMM-C the metadata had the following issue: /LocationKeywords/2/Type string "Continent > North America > United States Of America > California > NONE > Tuolumne River Basin" is too long (length: 95, maximum allowed: 80)</warnings></result>
From what I can tell, that means it's counting the full length of the keyword to include the Detailed Location, although I think strictly speaking, that's not part of the actual keyword. From what I understand, the "Tuolumne River Basin" part should be pulled off as "Detailed Location", an ancillary to the actual Keyword. Also, since the main keyword part ends with "NONE" at that point, it should be stripped off, leaving the actual, true keyword as "Continent > North America > United States of America > California" (after un-escaping the >'s of course).
Is there still something I'm overlooking, or is this still something that hasn't been rolled out in the CMR ingest code yet?