User:Reosarevok/Agenda for IA Meeting

From MusicBrainz Wiki
Jump to navigationJump to search

Archive / MusicBrainz meeting

  • Where: Internet Archive, 300 Funston, SF
  • When: October 22, 10:30am
  • Who: invited people only (Ian McEwen, Oliver Charles, Kuno Woudt, Rob Derwin, Alex VanValin, Robert Kaye & archive staff)

Cover Art Archive

There are a few improvements/problems we should address with respect to the Cover Art Archive.

Node Downtime

For the IA, keeping the service always online is less important than securing good archiving. Which makes sense, but uploading data and then have it fail frustrates our users. Avoiding downtime or managing it so that our community knows about current problems would be good.

IA will create a URL for us to ping that says that CAA is healthy or not. Samuel will get back to us. --RobertKaye (talk) 21:45, 22 October 2012 (UTC)

Artwork deletion

See http://tickets.musicbrainz.org/browse/MBS-4753. Every cover art uploaded to CAA stays forever on archive.org. This is problematic:

  • if a private photo is uploaded by error, it won't be deleted when edit is canceled or image removed from MB
  • archive.org is unnecessarily hosting questionable images that have been rejected by MB community (e.g. because of low quality or duplication).
  • release merges can often cause moves, and those seem not to delete the old version, so there's duplicated data

We will fix the issues that occur on our side, but empty buckets will remain on the archive. The archive will not delete empty buckets.


Missing thumbnails

http://tickets.musicbrainz.org/browse/CAA-23 --Nikki (talk) 15:48, 3 October 2012 (UTC)

Incorrectly rotated thumbnails

http://tickets.musicbrainz.org/browse/CAA-34 --Nikki (talk) 15:48, 3 October 2012 (UTC)

Reject unsupported formats

http://tickets.musicbrainz.org/browse/CAA-28 --Nikki (talk) 15:48, 3 October 2012 (UTC)

Include image dimensions

http://tickets.musicbrainz.org/browse/CAA-33 --Nikki (talk) 15:48, 3 October 2012 (UTC)

Content-Length invalid for HEAD requests for artwork

http://tickets.musicbrainz.org/browse/CAA-27 --Nikki (talk) 15:48, 3 October 2012 (UTC)

Deleting empty buckets

We've got lots of empty buckets that need deleting.

Wrong media type

items are being created with the wrong mediatype (data/other, when they should be image)

CORS header

The stuff we need for uploading multiple images, PNG support and better errors. --Nikki (talk) 15:48, 3 October 2012 (UTC)

CDN

http://tickets.musicbrainz.org/browse/CAA-26 : loading images could be faster.

Maximum image size

How To Add Cover Art claims the filesize limit is 15MB. However, we have 29 images larger than this, the largest of which is 33.5MB. If there is supposed to be a limit, then the IA haven't implemented it. (If there isn't, then we need to fix our documentation) --Nikki (talk) 17:40, 18 October 2012 (UTC)

There is no limit. We do need to fix our docs. --RobertKaye (talk) 21:34, 22 October 2012 (UTC)

GIF thumbnails

One of the dependencies for http://tickets.musicbrainz.org/browse/MBS-4114 --Nikki (talk) 17:40, 18 October 2012 (UTC)

SSL

Part of http://tickets.musicbrainz.org/browse/MBS-5339 -- the IA has some SSL certificates that would prevent us loading CAA artwork over SSL, and which should probably be fixed! Ianmcorvidae (talk) 13:45, 20 October 2012 (UTC)

We're going to use the /download endpoint -- this should reduce the number of redirects and allow SSL. --RobertKaye (talk) 21:58, 22 October 2012 (UTC)


Metadata matching

We should talk about how we can work together to match the music that the archive has against MusicBrainz and how to feed missing metadata to MusicBrainz. (ingrestr & matchr) How can we work together to leverage the music that archive is collecting?

Future projects

Some future projects for us to think about:

  • artist image archive
  • music archive access for research purposes