Thank you for your answer. I already went to the first article as that helped me to get as far as I described above. The second link explains adobe bridge use of metadata but is of little help in this specific case.
Parallel to my OP I've discussed this with a Lr plug-in developer of this plugin: Welcome - P2Lr: Picasa Import for Lightroom and he concluded (what I suspected) that this concerns a bug in Lr: Lr doesn't associate the XMP people metadata to the photo from which it came. He's working on a workaround, however I presume this should also be reported.
Hence a new question: where do I report bugs in Lr, or is this the right spot?
...this concerns a bug in Lr: Lr doesn't associate the XMP people metadata to the photo from which it came.
The cause of this issue is that Picasa and LR apply the industry's metadata standards differently to represent faces and people, but each method conforms with those standards. You can't call it a bug -- rather, the cause is that the standards don't specify one particular method. See this post for details: Re: Face image region data written according to XMP-mwg-rs, but NOT read by LR. It should be straightforward for a plugin to rewrite Picasa's metadata to be compatible with LR's representation.
It should be straightforward for a plugin to rewrite Picasa's metadata to be compatible with LR's representation.
I have read all the Lightroom SDK docs and done a bunch of experiments and cannot find a way to access the Region Name and XMP:Subject fields from within a Lr plugin. The Lr plugin API has a very restricted view of the metadata and only supports around a few dozen of them. If you are familiar with Lr plugin API and have a suggestion, I'd appreciate it.
I did successfully use exiftool before importing to copy Region Name to XMP:Subject as you suggested, and Lr imported them and linked the keywords to the photos (sort of... a little flakiness but its a long story). Thanks for that tip!
I would prefer to do it in my plugin but don't see a way to do it.
BTW, the shift-o edit trick described by wolph42 earlier in this discussion works as long as you don't have too many faces, in which case it chokes.
cannot find a way to access the Region Name and XMP:Subject fields from within a Lr plugin.
The plugin would have to invoke Exiftool to copy the region names into the XMP:Subject field. (Lots of plugins invoke Exiftool to access metadata not available via the SDK.) If you want to discuss this in detail, please post on the SDK forum.