AR don't appear in Picard for one (big) release

picard
tagging
ar
Tags: #<Tag:0x00007f2a0087f600> #<Tag:0x00007f2a0087f420> #<Tag:0x00007f2a0087f2b8>

#1

I’m trying to tag this release: https://musicbrainz.org/release/560856a8-3f9c-40a2-b366-09d394e5e16a and for some reason the AR are not saved in the tags. No composer nor performers. I checked both release relationships and track relationships in the options, and it works fine on other releases. Is it because there is too much information to load ? I uninstalled Picard and compiled the latest version (1.4.0dev5) but the problem is still there.


#2

That’s interesting, if one looks at a webservice request for this release, one can see that the result does not contain any relationship for recording or works:

https://beta.musicbrainz.org/ws/2/release/560856a8-3f9c-40a2-b366-09d394e5e16a?inc=release-groups+media+recordings+artists+artist-rels+release-rels+url-rels+recording-rels+work-rels+recording-level-rels+work-level-rels

The same request for another, much simpler release, does:

https://musicbrainz.org/ws/2/release/f2b69735-d954-4a27-a252-cfdb1b01aab4?inc=release-groups+media+recordings+artists+artist-rels+release-rels+url-rels+recording-rels+work-rels+recording-level-rels+work-level-rels

I think there was some server side change that prevents loading the relationships for such a massive release. Can somebody confirm that?

But actually without that change I would have expected the server to fail. We had this come up a couple of times, where the server failed to load a large release with too many relationships and just returned an error. If you got this issue in Picard the only workaround was to disable track relationships (or even release relationships).


#3

Yes, releases with more than 500 recordings are currently blocked from returning relationships: https://github.com/metabrainz/musicbrainz-server/commit/49cfb780de0bca2562a5ad14ecb8bab7e040a0c7

The request would’ve certainly failed (as in timed out because it would’ve been so slow) without that restriction, unfortunately.


#4

I have seen similar problems with large sets like this. I wonder if it would make sense to update the web services so that we can request only one medium at a time from Picard. It would be a logical way to break it up and some of these releases are gigantic!