A miscellaneous collection of bug fixes, and some back-end changes, and integrated Amazon cover art.
Changes mainly of interest to MusicBrainz Users
Amazon Cover Art
MusicBrainz now matches its database against the Amazon.com catalog.
For each match found we display the album cover art in the album detail
page. We’ve also added links for more information on the albums at
Amazon, and a “buy” link that lets you purchase the album from your
preferred Amazon store.
Associated with this are two new preferences: one lets you choose your
preferred Amazon store (e.g. the one closest to you); the other one
turns off both the cover art and the links to Amazon. One of the
conditions under which this content is licensed from Amazon is that we
must display the “buy” link whenever we display the cover art —
so don’t ask us to show cover art without the “buy” links, because we can’t
do that. It’s both, or neither.
Once we have a non-profit corporation for MusicBrainz, we will be able
start using our associate IDs and start earning a small percentage on albums
sold via MusicBrainz. In the process of adding the Amazon cover art,
we’ve also reworked the way that artist, album and moderation links are
shown to make the album page less cluttered.
These changes only affect “FreeDB moderations” — i.e. when
a web service client requests album information based on
a MusicBrainz disc ID (and TOC) that we’ve not seen before,
causing the corresponding album to be automatically imported from FreeDB.
Firstly, the code used to turn the TOC into a FreeDB ID was flawed;
sometimes it yielded the correct FreeDB ID, and sometimes it didn’t.
Of course, this has now been fixed. However it is likely that there
are now some albums which have been automatically imported into MusicBrainz
but which were then associated with the wrong disc ID and TOC.
It should be possible to write a “fix” script to identify and remove such
broken data; however that script so far has not been written.
Secondly, the minimum number of tracks need for an automatic import to
proceed has been raised from 2 to 5 (bug #800776).
Finally, the logic that tries to prevent the automated import of “Various
Artists” albums has been considerably strengthened; this should mean that
the FreeDB Moderations page will
tend to contain less junk in future.
Edit Artist Name bug fix
Previously, when an “Edit Artist Name” moderation was applied, the artist’s
page index (used for browsing) would be changed, when it should have
For example, if there was an artist “Bob Price”
(sortname: “Price, Bob”), then changing the name to “Bob Pryce” would have
moved that artist from the “P” page to the “B” page. Since the “Edit
Artist Sortname” moderation also (correctly) modified the page index value,
the page which Bob ended up on would depend on whether the most recently
applied moderation was the “Name” or the “Sortname” moderation.
This bug has now been fixed, but it is likely that there are still some artists in
the database who are on the wrong index page as a result. Editing
their sortnames will fix the error. At some point a “fix” script needs to
be written to find and correct such broken data.
Previously, by using the “I forgot my password” feature, it was possible to
discover the e-mail address of any user. This (bug #893529) was fixed
on the live server the day it was reported (so it falls between the previous
release and this one).
You’ll know if someone abused this flaw to find
your address, because you will have received the “this is your password”
reminder e-mail, even though you didn’t ask for it to be sent. As far
as I am aware, the only person this happened to was me, and that was because
the bug finder was testing
out the bug on me before he reported it 🙂
“Add Track” moderations on “Various Artists” albums previously showed a
blank artist name/sortname (on the voting pages, for example). From
now on newly entered “Add Track” moderations will show the correct artist;
moderations which were already in the database before this release will just
not show the artist name (bug #868238).
When selecting an “inline moderation” to display on a “Various Artists” page,
the server no longer first picks a moderation “by the same artist”,
since it is very unlikely that
the selected moderation will have anything to do with the page you’re
Many more timezones have been added to the allowed list, as seen on the
preferences page (bug #757534). Also a few extra
date/time display formats have been added.
You can now choose your preferred Google domain (e.g. http://www.google.de).
The MusicBrainz logo on the Google results pages is much nicer too 🙂
Some dates were being incorrectly displayed as being Jan 1st, 1970 (or
maybe Dec 31st, 1969, depending on your time zone). These dates
are now correctly displayed as “never”.
The cache expiration of the sidebar panel content (server stats, top
moderators, top voters) became broken during
the move to the new Mason.
It’s fixed now.
New and Changed Documentation
A couple of typos have been fixed on the MM page.
Another has been fixed on the mq_examples page.
Added the mb_server codebase.
Added release number / CVS Tag information to the
CVS modules page.
Changes mainly of interest to MusicBrainz Programmers
The database dumps
will now be performed every other day, instead of every day.
Changes mainly of interest to MusicBrainz Server Programmers
The MusicBrainz scripts now look for the Perl binary at
/home/httpd/musicbrainz/mb_server/cgi-bin/perl instead of
/usr/bin/perl. Thus, you can now easily run your mb_server
from one installation of Perl, but have your system-wide Perl (which maybe
your operating system depends on extensively) be kept separate.
All you’ll probably have to do is copy the Perl binary you were previously
using into the new path.
This change is described in the INSTALL file.
A cacheing layer has been added: MusicBrainz::Server::Cache.
Currently this is based on Cache::SizeAwareFileCache (from the
Cache::Cache distribution). There are new options in
DBDefs.pm to control this. In
the future the implementation of the cache might change; code using the
cache should neither care nor notice if it does.
So far the following things are cached in this way: user data
(UserStuff.pm); FreeDB responses; and user voting history statistics (as
displayed on the “profile” page).
Discid / FreeDB
As noted above, the FreeDB ID algorithm has been fixed — previously it
would sometimes give the wrong answer. In fact much of
have been tidied up and revamped:
Discid->GenerateAlbumFromDiscid has been tidied up,
as has Insert and InsertTOC.
Discid->GetAlbumAndIdFromDiscid has been removed.
Discid->ParseTOC has been added – this takes a TOC and returns
lots of useful things from it, including the track count, lengths,
MusicBrainz disc ID, and FreeDB ID.
In FreeDB.pm, _lba_to_msf and IsNumber have been removed,
and Lookup has been greatly simplified.
Finally, FreeDB.pm now caches query responses from the FreeDB server.
Other changes include:
XML::DOM is not required,
so it has been removed from the dependency list in the
A hack has been added to
to avoid re-executing the same SQL again and again for every page,
when it only needs to be done once (when the database connection is first
made). However this is a hack! — it’s not
done using proper DBI techniques, and it may break things, somewhere, some
day. Let us know if it causes you any trouble!
mq.pl now logs the authenticated user
(if known) into the access log. The URL logged is now of the form
mq_2_1.pl has been doing for a while
now. Also these two scripts have been tweaked in various small,
cosmetic ways so as to make them as similar as possible.
Bugs and RFEs Closed