Created attachment 317085 [details] [review]
libtracker-data: Silence a CRITICAL

I'm also curious where that NULL is coming from.

From #tracker on GIMPNet:

11:42 <rishi> ssam2: aleksander0m: garnacho: Hey! Does this make sense:         
      https://bugzilla.gnome.org/show_bug.cgi?id=759280 ?
11:47 <garnacho> rishi: I'm principle yes, I'm curious to which query triggers  
      this though
11:48 <garnacho> I guess you could receive nulls on things like . FILTER        
      (REGEX(nco:fullname(?u), ...)) if fullname is actually null
11:48 <rishi> Something from rygel. I can dig it out a bit later.
11:51 <garnacho> rishi: no worries, I think I can imagine other scenarios       
      where you may pass nothing there
12:09 <garnacho> rishi: and suspicions confirmed, please push :)
12:10 <rishi> garnacho: Ok.

FWIW, this may happen on queries where the regex value is unbound and not actually present for the element at hand, eg:

  select ?u { ?u a nco:Contact . FILTER (REGEX (nco:fullname(?u), "foo"))}

The sqlite engine would typically make this invisible (never pass nulls to regex()) if the query ensures the value is bound, eg:

  select ?u { ?u a nco:Contact ; nco:fullname ?fullname . FILTER (REGEX (?fullname, "foo"))}

Push to master and all the way down to tracker-1.2.

Comment on attachment 317085 [details] [review]
libtracker-data: Silence a CRITICAL

Forgot to update the status.

Could you track down where that was coming from in Rygel?