• 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Using MoSh for shared handbell scores
#1
Hi, all,

I play in a professional handbell ensemble (Sonos, www.sonos.org) and we play on a bunch of instruments (200+). We have 12 musicians who read off of 8 books, so most books are read by 2 musicians. Our scores are laid out using a typical grand staff and look like piano scores with extra symbols for handbell-specific articulations and expressions. We use all paper at this time. For background, most handbell groups use standard height tables with a layer of 3- or 4-inch foam rubber on which the instruments lay. We currently play with 34-ft of tables with instruments from C2 to E8 (handbells are octave transposing instruments). Each musician has their own personal copy of the music for individual practice as well as the books that we read from on the table, which means we have 20+ books all together.

We typically use a blue/red marking scheme for score annotations, where the player to the left of the book makes their annotations in blue pencile and the player to the right of the book uses red for theirs. Our music has been growing increasingly complex - our latest tackle is the Borodin Polovtsian Dances #17, which is extremely demanding and has many instruments loaded on the tables and thus we have a pretty significant number of markings in our scores.

This is becoming unmanageable and we've increasingly found it difficult to come to agreement on who is actually supposed to play what. So here's what I want to do:

1. Have a single cloud location for a score and all of its associated metadata (mainly markings organized as annotations).

2. The score will have annotations containing markings for each of the books on the table. When a piece is first being worked on, a leader (typically me) will do the thankless work of breaking down a full score into individual "parts" and then assigning those parts to each musician. (There is not a "best practice" for this strategy, but we use a time-honored tradition of assigning a consecutive range of notes to each musician and divvy up the entire range that way.)

3. Each book will have its own annotation layer, so I would create layers called "Book 1", "Book 2", etc. In those layers I would describe in text the overall breakdown (Blue (left) has this range of notes, red (right) has this range, etc.). I would also make additional markings to indicate changes from that overall breakdown in specific places, such as circling some notes or crossing out others in blue or red, depending.

4. Synchronize the shared "master" score to everybody's devices, including the tablets used on the tables (see what I did there?) and people's private devices for their own practice.

5. During rehearsals, musicians inevitably make changes to the initial markings I made. They'll make those changes in their own "book." These markings then need to be "uploaded" to the master so everybody can see them. The changes then need to be automatically synchronized to the tablets.

6. In this way everyone is always working from the latest, greatest set of markings.

7. During rehearsals early in the season, LOTS of changes are being made to markings by musicians all at the same time, so we need to be able to edit different annotation layers at the same time - can't have a "lock file; make changes; unlock file; resynchronize" gatekeeping system.

Crud, I know that's a lot of text. If you've gotten this far, thanks! I've read the MoSh manual and it seems to only support one-way synchronization, namely from a cloud directory down to devices, but I need two-way synchronization such that musicians can make changes on their individual devices and then "push" those changes up to the cloud directory. This is sort of like version control, I guess, and I imagine that it would work the same way - if two people are simultaneously pushing changes that don't "conflict" with each other, then no resolution is required and the cloud data is easily updated. For us the scope of change would always be in the annotation layers and two people would never try to change the same annotation layer at the same time.

Sorry again for the wall o' text and I hope that I can learn from y'all how to tackle this problem!

---Jason
Reply
#2
Unfortunately, I think there are some limitations in MobileSheets that are going to make it difficult to use for what you are trying to do:

1) Annotation synchronizations are pretty much "all or nothing". Annotations can't be merged - they can only be replaced, which is a problem if multiple users are making different annotations and they both need to push those changes. 
2) Synchronization is currently manual - users have to push changes, and pull changes. If multiple users synchronize to a cloud folder at the exact same time, there is no lock, but it could result in an unexpected outcome as the database is only uploaded at the end, so one user's synchronization would overwrite the others.

One of my goals is to support automatic synchronization of annotations when using the "Connect Tablets" feature. If you can have all the tablets connected over WiFi or bluetooth (bluetooth only supports 7 simultaneous connections), then that feature might be of use to you. However, if you need an offline way of synchronizing all the annotations, then I would have to support a concept of merging annotations, but this is really a can of worms, because there is no way to uniquely identify an annotation. While MobileSheets could try to make educated guesses to match up annotations, there are so many potential scenarios where it could get it wrong, and users would not be happy having duplicated annotations if it got it wrong. What I could do is start tracking annotation changes by layer though, and just replace individual layers based on when they were last modified. If each user worked on a different layer, than this would prevent a lot of conflicts and problems. I don't currently track the last modified timestamp for each layer though, and this is still potentially fragile. Some users have asked for the ability to pick which layers are synchronized, and while that's something I may consider supporting, I'm really not sure how great this will be in practice, because if the user has hundreds of layers across hundreds of pieces, that's going to be difficult to manage (that's assuming each layer had a unique name, as I would be matching on name). 

The last problem that would have to be resolved is the correct handling of multiple synchronizations occurring at the same time from different users. There is no lock currently, but I think I would need to implement one to ensure data integrity. However, if the synchronization occurs automatically in the background, it would need to be made intelligent so that it only synchronizes the small amount of data that has changed. If I can speed up the synchronization in this way, then the lock would be very short.

It may be some time before all of these issues are tackled though.

Mike
Reply




Users browsing this thread:
2 Guest(s)


  Theme © 2014 iAndrew  
Powered By MyBB, © 2002-2024 MyBB Group.