Major Improvements Approaching!
Happy November to all as football season reaches a homestretch. In case you haven't noticed, my own activity has slowed as of recently. This is primarily due to my work in assessing this website and its usage to plan out a massive site-wide update to cement ease of use, proper archival organization, long-time user suggestions, and solid navigability. I estimate that this update will be in progress for no longer than a week. Unlike others, however, it won't be shipped overnight and if you're a regular user of this site, there's a chance you'll catch things updating in real time. So this post will act as a prelude to these new additions and explain just about everything you need to know.
Why we stopped to look
The archive now holds records of about 10,000 published titles and many more still being processed, against 367 pieces of sheet music it physically holds and 172 photographs, recordings, programs and objects. The catalogue is the largest thing here by a factor of nearly thirty, and it got that way in about four months.
Almost nothing in the site is built around it. The front page does not mention it. No page lists it. The navigation has a link called "Sheet Music Catalog" that goes to a list of 74 publishers, which is a different thing entirely. A catalogue of ten thousand band publications with composers, arrangers and plate numbers does not exist anywhere else in this form, and here it is the hardest part of the site to find.
That one mismatch explains most of what felt awkward to me about using this place and it is what the next stretch of work is about.
The catalogue gets its own front door
A real page listing every entry across every publisher, with the filters this material actually needs: medium, publisher, composer, arranger, decade, grade, whether the archive holds the music or only knows the piece existed, whether there is a recording, whether anybody has recorded a plate number for it. The plate ordering that already exists on publisher pages comes with it.
The navigation link goes there, and the link to the publisher list goes back to being called Publishers, because that is what it is.
Search that shows you everything it found
Search currently returns six results per category and stops. There is no second page, no filter and no sort, so a search of ten thousand catalogue entries for "march" shows you six of them and a number. There is no way anywhere on the site to search the catalogue across publishers and page through the results.
Every category gets a "see all" that opens into a real list you can filter and sort. Collections and events get added to the index as well; both have their own sections and neither is searchable today, which means the arrangement layer for everything that is not sheet music cannot be searched at all.
The same filters wherever you are
The archive side can filter by date range and tag but cannot sort at all. The sheet music side can sort six ways but cannot filter by date or by tag, even though works carry tags and already have a year recorded. One set of filters, built once, so what you learn on one page still works on the next.
Part of this is less visible and more useful than it sounds. Items can record a date the archive only half knows, so 1987 with no month stays different from January 1987. Sheet music cannot do that yet, and it will.
One way to follow something, and somewhere to see what you followed
You can favorite a piece of sheet music. You can watch a catalogue entry. Those are the same gesture under two names, they work on different things, you cannot do either to the other, and nowhere on the site shows you either list back. The information has been collected and never returned.
One gesture, one name, working on anything with a page, and a Saved list on your dashboard.
Comments on everything that has a page
You can comment on items and on sheet music works. You cannot comment on a catalogue entry, which means the ten thousand records that make up most of the archive are precisely the ones you cannot leave a correction on. Folios, collections, events, organizations, artists and publishers support nothing at all, despite each having a section of their own.
Corrections and questions from people who were there are how a community archive gets accurate, so they should be possible on every record that has a page.
One place to start a contribution
There are five separate contribution forms, and you have to work out which one your material belongs in before the site will help you. That is backwards: sorting your own material into the archive's categories is the part you are least equipped to do, and it was the first thing being asked of you.
A single "add something" page will ask what you have in plain words and send you to the right form. The forms themselves stay as they are. They differ because the material differs, and a photograph genuinely does need different questions than a part book.
Holdings: what the archive has, and in what condition
The biggest new idea, and the one that changes what a record can say.
Right now a copy of a piece of music is close to a yes or no. Either the archive has it or it does not. What it cannot tell you is the thing anybody actually wants to know: is the set complete, which parts are missing, is this the original print or a photocopy somebody made in 1994, and who physically has it.
Holdings will record exactly that. Completeness and which parts are missing. Whether a copy is an original, a publisher reprint or a reproduction, because those are three different objects and calling all three "a copy" loses the distinction. Who holds it, whether that is the archive, a school still using it, or a private collection. And where a reproduction came from, so a photocopy can point at the original it was made from.
Two smaller things travel with it. Whoever contributed a holding stays credited on it, visibly. And the word "scan" gets used more carefully in the interface, because it currently describes both a digitized original and the physical thing, which are not the same claim.
Editions
Publishers reissue. The same piece comes back with a new cover, sometimes a new catalogue number, occasionally reset type, often printed from the same plates as the first run. At the moment the archive has no way to say that two records are the same piece in two printings, so they either sit as duplicates or get merged and lose the distinction.
One correction worth making publicly, since the last post talked up plate numbers: a plate number does not identify an edition. Publishers reprinted from the same plates for decades and changed everything around them. Plate numbers date a print and order a publisher's output, which is a lot, but editions need their own record and will get one.
This is the piece most likely to take longest, because what counts as "the same piece" is already used in several places on the site and all of them have to agree.
Availability, and what it is not
A noticeboard. Somebody who holds a set of parts can say they are willing to lend or sell it, and somebody looking for that piece can see who has it.
Stating plainly what this will not be: the archive will not handle money. No payments, no fees, no holding anybody's funds, not at any point. Becoming a payment facilitator carries obligations this project has no business taking on, and a noticeboard where people arrange things between themselves does the useful part without any of that.
It will also be deliberate rather than automatic. Nothing becomes available because of a setting somebody forgot about; somebody has to say so about a specific holding.
Magazines, catalogues and other things with issues
Publisher catalogues, band magazines and programs are already contributable, but there is no way to say which volume or issue one is, or to group a run of them together. Volume and issue numbers are coming, and a way to tie a run into a series, so a shelf of a magazine reads as a shelf rather than twenty unrelated documents.
Composer pages that are about the composer
An artist page currently gathers everything a person is credited on. For an arranger that is right. For a composer it is often wrong: a composer of a pop song that somebody else scored for marching band ends up with a page full of other people's arrangements of their tune, and their own original works are buried in it.
Those get separated, so a composer page leads with what they wrote and lists adaptations of their work beneath it.
Redacted copies
Some material can be shown in part but not in full. A redacted public copy standing beside a restricted original does not fit the current model, and the work to make it fit turned up one genuine leak worth closing while we are in there. Later in the order than the rest, and mentioned here because it is on the list rather than because it is close.
Organization pages that know what they hold
Some school pages say "0 items" when the archive very likely holds material from them.
Items record their school as text, and the links between items and real organization records were built later, so the catch-up has never been run. It will be, and then the text column can go, which stops the two from disagreeing in future.
About thirteen thousand more catalogue entries
A complete pass over one major publisher's concert band and marching band catalogue has been collected and is waiting to be imported: a little over thirteen thousand titles. About 650 of them match something already in the archive and are held back for review rather than imported blind, because overwriting a catalogued record with an external listing is worse than asking.
When that lands the catalogue roughly doubles, which is most of the reason the first two items in this post come first.
Work you will not see, which is the point
Two things that produce no new page at all.
Something like a plate number, one small optional field, currently has to be added by hand to about a dozen places: the forms, the importer, the record's own page, and every table anywhere on the site that lists that kind of record. There are ten such tables, each written separately. When one gets missed, what happens is that you notice and tell me, which is how most of the recent "this works here but not there" reports happened.
So those ten tables become one shared piece, and a test gets added that checks a field you typed into a form actually survives being saved. Three fields have quietly failed to save at some point for exactly that reason.
The second is a standing set of deliberately awkward records: a publisher credit nobody has confirmed yet, a piece with no publisher at all, a restricted copy seen by somebody not signed in. Four of the recent server errors, including the Incantation page and the PDF viewer, were all the same thing, a page meeting a record shaped slightly unusually. Keeping those odd records around permanently and checking every page against them closes that class of problem rather than fixing them one at a time.
None of this adds a feature. It is the difference between new things arriving finished and new things arriving and then being patched for three weeks.
Two honest notes
The archive has recently began keeping a log of what gets read. This is an adaptation of the "recent activity" section of this site designed solely for private, analytical purposes. Views of items, files, and search keywords are what gets recorded by this site. These records are accessible only by site administrators and are automatically cleared weekly by the site. Searches are logged whether or not you are signed in, because what people come looking for and fail to find is genuinely the most useful thing the archive can learn about itself, and was a major catalyst for the improvements you see listed here. That has been running for a few weeks and a plain statement of it on the site is owed and not yet written. It will be during the course of these updates.
Accessibility has not been looked at. Many have asked about what kinds of accessibility features this site will offer and I'd like to address this now as it has been something I haven't looked at during the course of building this site. As of now, this site received no keyboard navigation pass, no screen reader testing, and no contrast checking, ever. That is the largest real gap in all of this, it is not on the list above because the list came out of a different exercise, and it should happen before any visual work rather than after. After this batch of updates come in, focusing on accessibility will be a much smoother endeavor than it is currently.
The order
Roughly: the invisible groundwork first, then the catalogue front door, then search, then the shared filters. After that the following gesture and comments, then holdings and everything that depends on them.
The groundwork going first is not procrastination. Holdings alone add about eight new fields, and at a dozen hand-made places each that is a long tail of corrections arriving one report at a time over months. Doing the boring part first is what makes the rest of this list affordable.
As ever: if something here is wrong, or you were there and remember it differently, say so. That is what most of the above is for.