The same parser was used for both preferred content expressions and annex.largefiles. Reworked that today, splitting it into two distinct parsers. It doesn't make any sense to use terms like "standard" or "lackingcopies" in annex.largefiles, and such are now rejected.
That groundwork also let me add a feature that only makes sense for
annex.largefiles, and not for preferred content expressions: Matching by
mime type, such as mimetype=text/*
Yeah! I've waited for this option for a half a year
Thumbs up for finally including it!
Doesn't work out of the box, though; I do have the MagicMime build flag but it fails to add files now!
I get this error message for every file I try to add:
Don't know, unless your system somehow has the magic C library installed but no magic database file installed. You should file a bug report with details about your system and how you installed the magic C library.
For the simple reason that git-annex doesn't know the mime type of a file before it downloads a copy.
There are some more complicated reasons involving preferred content expressions needing to evaluate the same way in different repos, which can be on different computers with different versions of the magic database, etc.