Jaikoz- Discussion Forum

I haven't tried loading up any truely massive batches yet (largest recently was 13,000 files roughly) but I will soon. (relaized that on several batches I didn't run a discog check.)

But, in general things seem more stable and run more smoothly than ever before (now if we could wave a wand and improve the musicbrainz servers a bit...wish I had a local copy of it

Just eyeballing my memory I think I can start loading bigger batches.

mcammer, you mentioned you had an imac. Which processor is in your imac?

So Paul, you have a cross platform music tagging app. So do you have multiple machines/platforms for your testing before releasing it to the world at large or do you have trusted friends who test on their respective platforms and let you know of errors? Need anyone else?

Which is your everyday machine/platform?

While I assume you do Jaikoz development in the hopes of making mad cash, I have to assume it is also a labor of love because you love music. What kind of music do you find yourself drawn to (mostly)?

Ugh, sorry to bother you again, Paul.

2009..0324 4:29:27 PM [0x0-0x22022].com.jthink.jaikoz[185] 24/03/2009 16.29.27:com.jthink.jaikoz.monitoring.MemoryManager:addMemoryNotification:WARNING: Low memory limit is set for:Tenured Gen:1901454950

This was after rebooting the machine so almost no other memory was used.

Here's the memory line from my Info.plist

<string>-Xms500m -Xmx3200m -XX:MaxPermSize=150m</string>

I was only opening a folder of 30 items. I had only opened the folder and had performed no other operations.

Unfortunately this folder happens to have tons of dj sets and bootlegs and stuff so I wasn't too surprised that it wasn't finding stuff in musicbrainz.

Judging by console it appears to be a connection issue as I see the message

2009..0324 3:55:58 PM [0x0-0x7b07b].com.jthink.jaikoz[704] 24/03/2009 15.55.58:com.jthink.jaikoz.manipulate.musicbrainzhelper.MusicBrainzServerQuery:doQuery:WARNING: URL:Service Temporarily Unavailable:[700.7 700.0 10]

2009..0324 3:55:58 PM [0x0-0x7b07b].com.jthink.jaikoz[704] 24/03/2009 15.55.58:com.jthink.jaikoz.manipulate.musicbrainzhelper.MusicBrainzServerQuery:doQuery:WARNING: URL:http://musicbrainz.org/ws/1/track/?type=xml&limit=10&query=track:"we want your soul \(infusion mix\)" AND (artist:"adam freeland" AND qdur214 215 216 217 218 failed with error code:503

I know sometimes servers go down or are unavailable and it looks like it is trying the 5 times on each file before going on, but I just happened to have console.app up or I would never know and just assume musicbrainz had no info, not that maybe I should try again later.

paultaylor wrote:

Yes, I havent tried that release with Jaikoz yet, nor have I tried against the Java 6 release yet, but this is what you need to do,

Download Quaqua 5.1
Unzip then copy Quaqua.jar, libquaqua.jnilib & libquaqua64.jnilib to your Jaikoz.app/Contents/Resources/Java

I think that should fix it, let me know how you get on

Also looking console again, for everyone one of the originally posted errors/warnings there was this message :

2009..0324 3:13:15 PM [0x0-0x6a06a].com.jthink.jaikoz[612] Warning: SidebarTreeModel uses default user items.

After doing what you suggested both errors have gone away.

So I noticed both quaqua.jar and mac_widgets.jar in jaikoz. How's the transition?

Okay poking around the web, it looks like quaqua added support to their experimental 5.1 release on March 13th for the JVM SE6 (which is the JVM I am using).

but looking at release 5.0 their change log has:

Added libquaqua64.jnilib for use with 64-bit J2SE6 on Mac OS X 10.5.

So I am confused...

Poking around to investigate an out of memory message, I noticed this several times in my Console log.

2009..0324 1:54:51 PM [0x0-0x49049].com.jthink.jaikoz[389] Warning: class ch.randelshofer.quaqua.osx.Application couldn't load library "quaqua64". java.lang.UnsatisfiedLinkError: no quaqua64 in java.library.path

Does this cause any issues? Jaikoz still looks like an OSX app so is it reverting to a 32 bit library? Would using some 32bit libraries cause any issues if trying to use 64bit JVM and more than memory than 32bit JVM allows?

Can I download a 64 bit version of quaqua and stick it somewhere where it can be found?

Should I just turn off OSX look and feel in the preference?

I remember reading one of your blog posts that you were switching to MacWidgets library to get a more Leopard OSX look and feel. Has that been completed and is this an odd coding issue where something didn't get changed? (I see both mac_widgets.jar and quaqua.jar in the jaikoz.app package bundle)

Thank you for the quick reply, Paul. I believe I understand quite well now and hopefully this helped someone else. I shall bookmark this one so hopefully I won't have to bother you ever again about memory (unless I forget....that's a memory joke, folks!)

Thank you for adding the part about 64bit JVM. I was indeed already using in order :

Java SE 6 64bit
J2SE 5.0 64bit
J2SE 5.0 32bit
J2SE 1.4.2 32bit

Jaikoz doesn't appear to have any issues with SE 6. (Well maybe one which I will make another post about) except I think I see the GUI issue you refer to (I didn't even notice before).

Also in this thread (http://www.jthink.net/jaikozforum/posts/list/459.page), a user suggested turning off the display of the column for artwork.

Is this true? If so, does it still get updated even if not displayed? If so, I am currently displaying all fields, should I not display the fields I do not plan to actively edit and inspect in order to save memory?

Okay i downloaded v2.9 today and set it on a batch of about 3,500 files and fairly quickly got the dreaded out of memory message when "Correcting Tags from MusicBrainz".

As it had been awhile since gotten such messages, I double checked the forum and the help for instructions to increase memory. It looks like the installation of 2.9 overwrote my previous edits to Info.plist (I am on a Mac). The MaxPermSize is an option I am not familiar with from the last time I edited that file.

Paul, you have a thread here (http://www.jthink.net/jaikozforum/posts/list/776.page) where you describe increasing MaxPermSize but I will admit I do not understand how the three parameters relate to each other and to the system as a whole.

I am on a machine with 6GB of memory. When I use Jaikoz, I usually run only Jaikoz in order to ensure as much memory is available as possible.

So if I wanted Jaikoz to grab a big chunk of memory from the get go...

This is the line I have currently in my Info.plist :

<string>-Xms600m -Xmx1600m -XX:MaxPermSize=600m</string>

(Note: I multiplied everything by 4 before I decided to make this post and ask to ensure I understood the ramifications).

Xms : What is this? (starting memory?)
Xmx : What is this? (maximum allowed memory?)
MaxPermSize : It looks like this is related to Xms. How is this different than Xmx?

Do these values need to change in a relationship/ratio to the other parameters or can they be changed independently from each other?

If I understand correctly, Jaikoz gets a full allocation of memory upon launch that it cannot increase. True? (doesn't seem like it is true, because of the Xms and Xmx) If true, then if I wanted Jaikoz to use as much memory as it can grab (aside from the default system) which value should change?

cool. I'll give this a try after jaikoz finishes the batch it is on.

You can set certain fields to "never populate" to preserve your previously entered tags. Or you can set to "only populate if empty".

As for genres, i don't really have an opinion as genre is pretty subjective.

personally, I try to batch in groups of roughly 3000-5000 files.

Why? easier to eyeball and spot check, I'm not waiting so long for results (I like periodic gratification), and allows for some stopping points in case I decide I need to do something else.

Plus after doing a batch of about 12,000 files, I got the dreaded "low memory" message part way through.

If you set up and customize your autocorrector tasks, you can have it automatically save at multiple points if you like. (btw, Paul, thanks for adding this task to the list of options) This way, if you set up a large batch and encounter any difficulty, your previous steps are committed. I find there's no need to save after every step, however, as some take relatively little time and should I need to redo that step for some reason, it's no big deal (of course, on 95,000 files even the short time tasks could take awhile).

As I was describing in another thread, I have my autocorrector set up like such :

The various local correct tasks (basically use the default order and just append more tasks to the end)
Save files
Retrieve acoustic ids (there is a preference to save them to the files as they are retrieved so no need to save again)
Save (unless preference described above is set)
Correct tags from musicbrainz
cluster albums
Update tags from discogs
Save files
correct lyrics
save files
fix subfolder via tags
fix filesname via tags
save

Thanks very much

paultaylor wrote:

wynlyndd wrote:

Any benefit to having this both before looking up data on music brainz and afterwards?

No, only havbe it afterwards because it looks for tracks that share the release name but have a different release id to other tracks, and tries to minimize the number of release ids that a release is spread over.

So cluster albums after musicbrains look up...does the clustering merely change the release id or does it change the release id and other associated tags or should there be another lookup to musicbrainz to correct tags?

So would an idealized autocorrector would be...

the various Local corrects
Save
Retrieve acoustic ids
Save
Autocorrect from music brains.
Cluster albums
Update from discogs
Correct lyrics
save
fix subfolder from tags
fix filenames from tags
save