I've read this a couple of times:

All songs must be:
(
  (
    30% rating greater than 3 stars
    30% rating greater than 4 stars
    60% playcount equals 0
  )
  and
  Unique
  and
  Not: genre equals audiobook
)

and it still doesn't make sense....

>From a layman's point of view, I'd expect the % values in brackets to add up
to 100%. Because my understanding would be that 30% of my songs should be
above 3 stars, 30% above 4 starts and *40%* never played before
(playcount=0)

With this:


  Not:
  (
    60% a random song
    40% artist name equals balmoreha
  )


again, I don't understand what the "Not:" does to that whole query...

Thirdly, I'm sure there's a reason why the 1.4 UI was rejected, but is there
anything that that UI couldn't do? It was simple, but created very specific,
predictable playlists. I've tried this branch, and I got completely
confused! :-)

Peter

"Happiness makes up in height what it lacks in length"
-- Robert Frost


On 21 February 2011 02:24, Lukas <1luk...@gmail.com> wrote:

> If I might put my 2 cents...
>
> I think the key reason of the problem could be rephrased into:
> *Its is being done inside out*
>
> The problem is that user has FIRST select bias,  LATER put the settings in
> - like click plus sign and add artist bias, then type in artist name, then
> set the proportion then repeat all again for each other bias. Its just too
> much (in most cases) clicks required to even start doing it :(
>
> But if it would be possible to FIRST have search box looking for any meta
> filed with AUTO COMPLETION. Once user finds the keyword, like U2, it gets
> dead easy *for Amarok to know*, that U2 is an artist. So by simply typing
> in U2 and clicking enter new bias of *Match artist name: U2* can be
> created.
>
> Once done, put focus to search field again, so *U2 <enter> sting <enter>
> Gorillaz <enter>*  creates the playlist of U2, Sting and Gorillaz. Tick
> the "and similar" check box and playlist gets filled with by track of U2,
> Sting, Gorillaz and similar to the artists.
>
> *The end.*
>
> One more thing. Biases suppose to be grouped. I can't imagine any reason
> why user could want to have 3 blocks for each artist, instead of one block
> for all of them.
> The issue could be level of proportion. But 90% of cases it is no issue.
> Just do
> SELECT * FROM `music` WHERE `artist` IN (u2, sting, gorillaz) ORDER BY rand
> LIMIT 15
>
> For the remaining 10% uses, a something like its done in some players with
> volume control - an 22x22 icon on the right to call horizontal slider to set
> it up.
> Or even simpler way - just add a star. Once artist is stared in the bias,
> it forces the stared artist to appear much more fraudulently. Even 10 year
> old could do that.
>
>
> As I wrote about artists only, same way can be used for almost any other
> meta field.
>
> If there are any biases that cant be done in simple search > do way, they
> can be presented in the same way as it is now, just put them on the bottom
> of the panel. This way they wouldn't bother daily users, yet leaving exactly
> same functionality for power users.
>
>
> P.S. what about range slider http://jqueryui.com/demos/slider/#range to
> get used for range selection like rating? (both edit filer and biases)
>
>
> _______________________________________________
> Amarok-devel mailing list
> Amarok-devel@kde.org
> https://mail.kde.org/mailman/listinfo/amarok-devel
>
>
_______________________________________________
Amarok-devel mailing list
Amarok-devel@kde.org
https://mail.kde.org/mailman/listinfo/amarok-devel

Reply via email to