MediaMonkey forum

No, I don't think that it has something to do with the percentages as it wouldn't result in an overflow. And I treid some different cases out. It really seems to be a problem in the formula. I'm investigating...

@BerniPi: Any idea?

Puhh, I'm quite happy that I released some beta version first.

I have all the 1/2 percentages set to 0 and I do not get that error.

Nyn

Next try: 2.1 RC6

Changes since RC5:
-Changed: Improved the handling if sum of percentages isn't 100%
-Changed: Reverted to formula introduced in 2.1 RC1
-Fixed: Minor bug

I reverted the formula because the one I implemented in 2.1 RC1 (which was also new) was working better in some situations than the one implemented in 2.1 RC4. There wasn't only an overflow-problem but also the SkipCounter had too much influece. Example: A song with a playcount of 200 and 5 skips can get the lowest possible rating.

Berny:

Is there a way when implementing a new version that the new version doesn't change the data for the previous ratings from AR to MR? I would like to try to keep track of this data but how can I tell what is truly manual ratings if this keeps happening?

Nyn

Hello,

i actually knew, that the skipcounter isn't accurate in my formula.
Maybe this version is (subjectively) accurate:

10000000000 * (log(Played+1.01) - log(Skip+1.01)^2) / DaysSinceAdded^0.125 / 1.001^(1 / log(Played+1.01) * DaysSinceLastPlayed^1.1)

furthermore, i don't know, where the overflows happen. maybe some statistic s from your database would be usefull (Maximum skipcount, playcount, earliest dateadded, ...)

maybe there was a problem with the skipcount^2... (it got multiplicated with 10^10)

please try the formula and when there are problems reduce the "offset" of 10000000000 by some zeros.

nynaevelan wrote:Is there a way when implementing a new version that the new version doesn't change the data for the previous ratings from AR to MR? I would like to try to keep track of this data but how can I tell what is truly manual ratings if this keeps happening?
Nyn

This doesn't happen here and shouldn't happen! Are you really sure? So all your autorated songs have a 'MR' in the customfield?

Anyone else with that problem?

Big_Berny wrote:

nynaevelan wrote:Is there a way when implementing a new version that the new version doesn't change the data for the previous ratings from AR to MR? I would like to try to keep track of this data but how can I tell what is truly manual ratings if this keeps happening?
Nyn

This doesn't happen here and shouldn't happen! Are you really sure? So all your autorated songs have a 'MR' in the customfield?

Anyone else with that problem?

Yep, really sure, I went from a couple thousand with AR to all with MR. Just erased them all and am in the process of recalculating. I will keep an eye on this and let you know if it happens again because I do not know which RC version made the change but I know it happened before #5.

Nyn

@Nyn: But now it seems to work, right?

I won't know until I rerun the AR after an update, it doesn't happen in the same release, it seems to happen the first calculation after an update. Any new updates soon??

Nyn

Thanks Big Berny! RC 6 seems to do the trick. I just can't figure out why the RC 5 formula worked on some systems and not others (namely mine).

Sorry for the double post, but I am curious to know if everyone is running 3.0.3 Alpha 1? Maybe it is an error somehow created by the latest MM build.

Balthazar2k4 wrote:Sorry for the double post, but I am curious to know if everyone is running 3.0.3 Alpha 1? Maybe it is an error somehow created by the latest MM build.

I´m running it. RC6 works now.

----------------
Now playing: Drive-By Truckers - The Righteous Path
via FoxyTunes

@nyn: This will probably be the last release. At least I hope so. Just install the RC6 again over RC6. It will be the same situation like after an update. Would be cool if you could report what happens! Thanks for your help!

@Balthazar2k4: It's normal that the formula can cause problems on some systems while it runs ok on others. It happens for example when someone has high Playcounts or Skips. Then the script get's to high values when calculating the points using the formula which results in an overflow. The same formula can work nice if the PlayCounter and Skip aren't so high.

Btw, I'm using 3.0.3 Alpha 1 and it works perfectly.

Big_Berny

Big_Berny wrote:

nynaevelan wrote:Is there a way when implementing a new version that the new version doesn't change the data for the previous ratings from AR to MR? I would like to try to keep track of this data but how can I tell what is truly manual ratings if this keeps happening?
Nyn

This doesn't happen here and shouldn't happen! Are you really sure? So all your autorated songs have a 'MR' in the customfield?

Anyone else with that problem?

I can tell you that I have manually rated NOTHING in my library and I have many tracks with an MR code. In fact, it looks like most of them are tagged with MR. I just made an autoplaylist to filter them and 11340 out of ~14000 tracks have an MR code. I'd guess the difference is the same as the number of tracks I've played since I first installed your script. I'm not sure when or how this bug was introduced. It appears as if everything that had a BR at one time has an MR now. All the tracks in the list have 2.5 stars, my Base Rating.

I've just tested a few tracks which have an MR rating and I can say for sure that your Don't overwrite manual ratings option is working! LOL And here I was wondering why a lot of stuff I was playing wasn't rating automatically. I've just unchecked it and now stuff is getting a rating.

That looks like a bad bug, if you can figure it out.

I don't think it's that bad. Could you try the enw version and do an autorating, probably it gets fixed: http://www.berneker.ch/AutoRateAccurate21RC7.mmip

I had to remove something I implemented before for Spazz...