Abstract idea of Song instead of pure file management

[Jud Craft]

My idea was that you could match songs based on tags alone.  If two
songs of different formats have the same title, album, and artist, I'd
group them under the same entry in the database.  Saves all the
trouble of audio analysis -- which seems like overkill, just to map
different files to the same song information.  That's like killing a
squirrel with a mortar cannon.

Tag-matching alone would be prone to errors if the tags mismatch (a
definite problem), yes -- but if the user has two copies of the same
song in different formats, odds are that user intentionally put the
song into two formats.  If so, I think they can likewise be entrusted
with making sure they at least share common tags.

In addition, even with tag errors taken into account, disregarding
tags entirely also ignores the common-sense scenario:  if I have two
different-format files with identical tags, they should be mapped to
the same song, without question.  No analysis needed to figure that
one out.

I think it's a logical outcome of tag management -- if you're trusting
the user to have proper artist and album tags (and it appears you are,
otherwise the Amarok database wouldn't group by those) -- then you
might as well assume they can get the Track Name tag right as well,
and if that's the case, you can depend on those for easy song-file
matching.

I understand the song analysis part is useful for files that may have
mismatching metadata -- and I think song analysis IS a cool feature,
with many other uses besides -- but don't forget the simplest scenario
here:  if the files have the same tags, they should be grouped
together.

I'm glad to hear everyone has been planning for something like this though.