Composers, Tracklists and Performances, and a Long List of Fixes

Arrangers have always been real records here with a name you could click, with everything they scored gathered on one page. Composers were only text. So an original work, or anything nobody re-scored, was credited to nobody in particular. That is most of this catalogue: roughly 7,500 composer credits against 2,500 arranger ones.

That has changed. There is also a fortnight of other work in here that has not been written up yet.

Arrangers is now Artists

The section at Browse → Artists now holds everyone the archive credits on a piece, arrangers and composers together. Old /arrangers links still work and land on the same person, so nothing you have saved or shared is broken.

A piece is still credited the way a program would credit it: the arrangers when it has any, the composers otherwise. So a marching band chart of a pop song still reads "arr. So-and-so", and an original work finally credits the person who wrote it.

One deliberate omission. A composer named on a chart only because they wrote the underlying song. Beethoven under a band setting, say, appears on that piece but is not listed among the artists. That list is meant to be the people who made what the archive holds, and padding it with every songwriter in history would bury them.

A piece can name more than one composer

The contribution forms take composers in separate boxes now, the way they already took arrangers. Before, two people had to be typed as one string, which made "Rodgers & Hammerstein" a single person with a single page.

Recordings have tracklists

A record, tape or video can list what is actually on it: sides, track numbers, and who is playing on each one. Where a track names a piece the archive can identify, it links straight to it.

Where it does not, ticking a box files that track into the catalogue as a documented piece. That is the honest statement, because holding a record that names a piece is not the same as holding the music. It shows as awaiting a contribution, and becomes a full entry the moment somebody adds the sheet music.

Pasted links play, too. A YouTube URL attached to a recording used to sit there as plain text.

Performances have a section of their own

/performances collects every recording the archive can show, filtered by organization, decade and kind. Undated material sorts last, because a wall of "no date" is a poor way to open a section about when things happened.

Adding one is quicker than it was. /contribute/performance asks for the link, a rough date and who is playing, and nothing else. Almost every performance worth cataloguing now is an upload where format, condition and shelf location simply do not apply, and being asked for them was enough friction that recordings were not getting added.

Dates can be partial. "1987" means the year is known and the month is not, which is a different claim from January.

An occasion, such as a contest or a parade, can group several recordings together. You file a recording into one from the recording's own page, which is where you tend to notice that two videos are from the same afternoon.

Organizations can be built out

A school, corps or college program can now be created and edited, rather than only existing because somebody typed its name onto a contribution. Anyone signed in can add one. Editing is open to moderators and to whoever added it, since a page describing a real body is shared data rather than one person's.

They can also be connected to each other: a band that was succeeded by another, or two that are simply related. The connection is stored once and reads correctly from both ends, so the two pages cannot drift apart.

The catalogue is fully machine-readable

/catalog.txt and /catalog.json now carry every field the archive records for a piece, including publication dates, notes and the source each entry came from. Filters and paging are documented in the header of the text version. If you are doing research on this collection, that file is the intended way in. You should not have to scrape pages if you're looking towards ease of access for research purposes.

Smaller things

• The copy type "Full score" is now "Score & parts". Some people were choosing it for conductor's scores, which is a different thing and sits right below it in the list.
• Large files now go from your browser straight to storage instead of through the site, so a big scan or a fifty-file batch is no longer limited by what one request can carry. A refused upload also says why it was refused, instead of quietly reloading an empty form.
• Every item type can name a school. A concert program, the type most obviously for a school, could not before.
• Cover images and extra pages are offered for everything except photographs, where the item already is the picture.
• Moderators merging a duplicated name now type the correct spelling instead of picking from a list of every artist on the site.
• Spreadsheet imports of catalogue data now say what should happen to each row: create, update, skip or delete. A large one runs in pieces, so it cannot outlive the connection halfway through.

About the outages

Some of the above did not land gracefully, and if you were here over the last two days you saw it.

The artists list and the catalogue file were slow, and for a stretch they timed out entirely. Crediting composers multiplied the size of the tables behind those pages, and the queries reading them had not been rebuilt for it. As a result, the database was reading far more than it needed on every page load. That is fixed, and those pages are quick again.

Separately, submitting sheet music with a scan or cover image attached failed. Uploads now go from your browser straight to storage rather than through the site, which is what lets a large scan through at all. However, the new path had two faults that only existed on the live site and not in testing. Both are fixed. If a submission of yours failed in that window, nothing was half-saved; please try it again.

Still to come

A pass over the credit data itself is underway. Some names arrived from library and copyright records with fragments attached, such as a stray date, a bracketed correction, the word "arr" or other descriptive information where it's not supposed to be, and a few pieces that credit the right person in the wrong field. Those are being corrected against the original records rather than guessed at, so a few hundred names will tidy themselves up over the coming days.