I’ve started to experiment with PhotoMechanic (V.5) again, after moving away from Lightroom. I notice that it creates its own .xmp for each ingested image. Can darktable make use of any of the info. in this .xmp? If so. how?
You’d need to share an xmp file from photo mechanic with us, or turn on “look for updated xmp files” option in darktable and try it yourself. Or consult with photo mechanic’s support.
This is a dummy example of my playing about in Photo Mechanic210124_Snow-9Frx_0743.XMP (2.2 KB)
Likely not. It is like expecting 2 Anthony’s are compatible. 
In my experience (more than 7 decades), even 1 Anthony are incompatible…
Don’t get me wrong. Yours is still an interesting question. You likely would need to do some conversion or translation. I think it is easier between open source software however. Also:
Prior to founding Camera Bits, Dennis worked in the field of neural networks
Not sure if Photo Mechanic does but dt doesn’t use machine learning as far as I know.
And I thought being old… 
I’d expect that Rating, Subject, Description, Creator, and Rights show correctly. The rest of it is in IPTC or photoshop namespaces, which I don’t think darktable reads.
I can understand (I think) why dt would not bother to read variables from photoshop namespaces, but why not those from IPTC (really IPTC IIM)? After-all, if I have read correctly, XMP, as a standard, has largely superseded IPTC IIM but there is synchronisation between the two.
A really useful stock phrase for use in one’s CV. I, myself, used to work in the field of Lunar Exploration (picking up sticks, mostly).
Sorry about that; I’ll get my coat; I was just leaving anyway…
Who did you play with? 
Ha! That’s far too easy: have a go at really picking up sticks in the Australian outback when as a smart-arsed city kid you have no idea why you’re doing what you’re doing, you’re dying of thirst when all the proper workers around you think it’s just a nice day and you’re realising those Taipans don’t mess about…