TrackLength is seconds rather than milliseconds

[Jeff Mitchell]

Mark Kretschmann wrote:
> I think what you're seeing there is another symptom of our botched
> database update code, which was necessary because of the big "seconds
> -> milliseconds" migration we did all over Amarok. This bug is very
> serious, but we're aware of it. That said, we really need to come up
> with some ideas ASAP for fixing it somehow (I'm currently tending
> toward forcing a full collection rescan).

One thing to think about is whether we want to merge my uidhash scanning
branch in for 2.2.1. The problems people have seen were minor bugs here
and there in my SQL generation; the logic overall seems to be sound. The
only bug I currently know of is that opening the Tag Dialog on
individual tracks from the collection browser has a problem, but this
could get worked through. (I'm not sure if there are any other places
using SqlQueryMaker  that might have issues though.)

If we do, it forces a migration to using utf8_bin collation, which
requires a rescan -- so it'll be required anyways.

Unless we give it thorough testing, both in the database updating code
and everything else, I'd recommend it wait until 2.2.2, but it'd kill
two birds with one stone since it'd mean one (currently scheduled) full
rescan instead of two.

--Jeff

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 261 bytes
Desc: OpenPGP digital signature
Url : http://mail.kde.org/pipermail/amarok-devel/attachments/20091026/0c2dc3f3/attachment.sig