[All] To build a better music player

Stefan Monnier monnier at iro.umontreal.ca
Wed Mar 4 18:06:41 CET 2009

> I think it's a problem of horsepower.

I don't.

> If you can, as modern players do, read  all id3 info of all tracks,
> and process it on a relational database, you can  perfectly sort all
> your songs by author/album/year/genre/what you want ;)

Actually, if you only use ID3 info, it's very difficult to do a good job
of sorting files: it's basically impossible to reliably figure out which
songs are part of the same album and which aren't.  You can use
heuristics which will work OK in many cases, but sooner or later you'll
bump into some songs whose "album" tag says (say) "Anthology" and which
are really divided into 4 different albums, some of which are "well
behaved" (same artist and year for all their songs), and the others are
compilations where every song has a different year/genre/artist.
And then you add classical music into the mix and it gets even more fun.

> But, if doing so is a slow process on my 2.4 GHz desktop, I cannot
> imagine how slow would be on the freerunner.

My measly WL-700gE (266MHz MIPS CPU + 64MB RAM) does it just fine for my
music collection (7000 songs).

Whoever has read my earlier messages knows that this is using MPD ;-).

So at least in the case where the backend is MPD, pythm does have all
the necessary tag info at hand, but sorting is still a nightmare, so
sorting "by file name" really turns out to be a very good solution.


More information about the community mailing list