@ Democrite: I strongly disagree with your line of reasoning. Please let's remember that it's child-easy in Marvin to fix your meta-tags (including author meta-tags) if they are incorrect.
So just because many books have incorrect meta-tags, you're going to cancel a Marvin 2 feature, and will dumb down Marvin 3 so that it sorts authors in the crudest way imaginable – according to their first names?
I surely hope not.
If a user has chaotic meta-tags in his or her library and then complains to Kris that Marvin sorts authors "incorrectly", the answer on Kris's part would be very clear and conveniently invariable: "Please fix your meta-tags, and then your sorting of authors in Marvin will be perfectly fine."
Please look at Calibre: does it "baby" its users in terms of sorting in any way? Not at all. If the user's meta-tags in Calibre are incorrect, then the book titles/author names in Calibre will be sorted incorrectly. No apologies on Calibre's part – and that's exactly the correct attitude and approach to sorting issues that also Marvin 3 should adopt.
@ Chrisridd: I'm not sure I understand your question, but in any case, my answer to you is the same as to Democrite above: Marvin should respect whatever meta-tags the user feeds it. The GIGO principle is all-powerful, and applies here as well: "Garbage (meta-tags) In, Garbage (sorting) Out." I have had no trouble whatsoever in sorting books with multiple authors in Marvin 2, because Marvin 2 consistently respects the meta-tags as set in the (albeit misleadingly labeled) "Author sort" field. Yet Marvin 3 does
not consistently respect the data in that field, and that is – as I feel – an undeniable and unacceptable failure on Marvin 3's part.
