My experience with "random mix" has also been that it seems to get
stuck.
But after all random is random.
Fortunately there have been some cool plugins written to help with this
problem. The "SQL playlist" plugin seems to help alot. Some others are
Dynamic Playlist, TracKStat, and Custom Skip.
Could you retag Marillion as it's own genre and then exclude it from the
randomization (if you want)?
--
jbraveman
jbraveman's Profile: http://forums.slimdevices.com/member.php?userid=16943
View this thread: http://forums.
Doing a search through the 7.3 and 7.4 checkins emails, these are the
only change that I could find in the last couple of months. The
changes were made to the 7.3 branch, then merged into 7.4. Nothing has
changed recently.
December 9, r24559 (SC 7.3 Trunk):
> Bug: 8201
> Description: reset gen
I need to run a little longer, but I ran updates this morning on both
SqueezeCenter (7.4-24847) and MySQL (5.0.51a-23, debian unstable) and
the behavior seems to have gotten better. It's not pulling 2-3 songs
from the same album in the same random mix. It very well could have
been an issue in th
mrfantasy;391968 Wrote:
> And actually, you're oversimplifying--you're assuming you have 365
> artists in a uniform distribution. I'd expect about 15% of any random
> mix to be one artist (Marillion) since they're about 15% of my track
> count. The combinatorics are actually more complicated th
And actually, you're oversimplifying--you're assuming you have 365
artists in a uniform distribution. I'd expect about 15% of any random
mix to be one artist (Marillion) since they're about 15% of my track
count. The combinatorics are actually more complicated than that (and
they are with the bi
That all sounds reasonable, and superficially the data I have reflects
this, but I'm getting the same few artists and albums over and over,
not just the same one twice here and there. And yes, I know the
temptation to find patterns where none exist; so I will try and pin
this down more quantitati
(and now that I have had my caffeine..)
Since I am too lazy to do the math, we'll pretend you have 365 artists
in your library.
If you choose 23 random tracks, then, you have a 50% or so chance that
two of the tracks will be by the same artist.
If you choose 30 random tracks, then about 70% of
mrfantasy;391479 Wrote:
> It's the Birthday Paradox that would make the fact that 10,000 "random"
> tracks from my collection have no duplicates impossible without a rule
> filtering them out.
I said that when you mentioned the Unique requirement in your 10k item
test, that it reminded me of the
Philip Meyer;391519 Wrote:
> Another way to check is to turn on SQL query trace logging to find the
> SQL query that is being executed to get the list of random songs.
> Check the SQL command is correct, then check that the songs being
> played match the songs returned from the command.
>
> Phi
I'm not using 7.4 yet but I have had the same feeling in the past and I
have a similar size library. I've usually found that a clear and rescan
fixes things but whether it fixes SC or me is unproven :-)
Craig
--
Craig
MC2Slim - Windows Shell and J River Media Center Integration for
Squeezebox
Another way to check is to turn on SQL query trace logging to find the SQL
query that is being executed to get the list of random songs. Check the SQL
command is correct, then check that the songs being played match the songs
returned from the command.
Phil
The way to test this, I think, is to set a random song mix, then save
the playlist, then set a new mix, etc.
Save a bunch (say 10% of your music collection) and see if the
frequency of artists/songs/albums is truly random and representative of
your music collection. I think that setting larger m
snarlydwarf;391428 Wrote:
> Are you sure it's not a variation of the Birthday Paradox?
>
> Expanding the limit to 10k, but keeping the unique is why this reminds
> me of the Birthday paradox.
It's the Birthday Paradox that would make the fact that 10,000 "random"
tracks from my collection have
Are you sure it's not a variation of the Birthday Paradox?
Expanding the limit to 10k, but keeping the unique is why this reminds
me of the Birthday paradox.
--
snarlydwarf
snarlydwarf's Profile: http://forums.slimdevices
I think that I agree. This is difficult to prove, but some artists or
albums seem to get a better chance of appearing than others. Does
anyone have an suggestion about how to test this?
BTW, I am using SC under Fedora 10, in case that makes a difference.
--
EelcoV
-
The "clear and rescan" hasn't changed the behavior--it's still pulling
from the same list of 40-50 artists and compilations (out of 666
artists).
I set Random Mix to select 300 songs, so it should look like a wide
variety of my music collection (13311 songs). I just did 1 songs,
and it did b
I know this gets into descriptions of what truly is "random", but I
don't think Random Mix in the latest 7.4 builds is doing it for me.
I think it's only getting a subset of my actual music--a certain list
of artists or albums instead of the whole thing. It's hard to see
because it also seems to
18 matches
Mail list logo
