No materials for the event yet, sorry!

Snap! is available in 47 languages, and has a truly international community of users. The architecture that has made that possible, and the efforts of the folks who have helped with those translations, are deeply appreciated.

However, having worked for the last two years to translate the Beauty and Joy of Computing (BJC) AP CS Principles curriculum to Spanish, we have discovered that what might seem like a simple task of moving a curriculum to another language, revealed a remarkable list of inter-dependencies on English and difficulties in managing that large project. So many tasks were labor-intensive, and scripting didn't help. Among the challenges:

• There was no automated way to translate all the hundreds of PNG images of blocks. “Smart PNGs” (with the XML embedded in the metadata) is amazing and now allows for automation, but the images were created before that feature existed. Animated gifs showing how to do something had to be recreated by hand.
• Not all the blocks (or error messages, or menu items, or dialog boxes) in Snap! have been translated to all 47 languages, since once a translation is created, and new features are added, all translation teams need to be informed about the additions. Perhaps a team needs to realize it's not a one-and-done process and check back in every 6 months?
• Snap! help text currently only exists in English
• A curriculum translation often takes a fair amount of time, and sometimes the curriculum or Snap! changes before it's done, so managing those different versions can be difficult.

Could automation and large language models help? Should there be a “translation regression test” feature? It would go through all the blocks, libraries, menus, dialogs, help screens, etc. and create a PDF with three columns: the English on the left, the translation on the right, and the place in the translation file to make the change.

These are just some ideas we'd like to explore in this BoF. If you have been involved in any translations of Snap! (or curriculum built on it), please share best (and worst) practices. If you have ideas to leverage automation or tooling to help with future translations, please attend and share your thoughts!