On Thu, 2005-04-28 at 09:25 +1000, Jonathan Matthew wrote: > On Tue, Apr 26, 2005 at 11:04:53PM +1000, James Livingston wrote: > > One word of warning: if you have a playlist that uses either of these > > two criteria, and try to open an older version that doesn't support them > > _IT WILL CRASH_ as older versions don't understand my new > > "current-time-within" operator. > > Does anyone think it's worth handling this a bit better? Maybe show > the playlist as broken or disabled, and somehow manage to write the > query out to the playlist file untouched? That sounds like a good plan. It won't help with all the existing copies of RB out there that won't be able to understand playlists with new features, but it will stop the problem happening again. To actually make 0.8 not explode would require not inventing any new operators/properties/whatever else it doesn't like. This would mean overloading less-than and greater-than for the current-time-based properties, which is doable, but I don't know if it is a good idea. > The attached patch fixes a bug where you'd get an assertion failure if > you had text in the search box and you selected an artist/album/genre > from the browser. The resulting query has 3 components (song type, > search text, artist/album/genre), but rhythmbox_query_magic_transform > was only handling queries with 2 components. Thanks, it was something I forgot to remove when I "borrowed" some of the code from over in the playlist gui. After looking at it I'd better apply your patch too - Oliver's works fine, but yours probably make it more "future-proof". James "Doc" Livingston -- People use dummies for crash-tests. Windows is so difficult they had to educate the dummies first -- by giving them "Windows for Dummies" books! -- Ewout Stam
Attachment:
signature.asc
Description: This is a digitally signed message part