I understand that S3 is seen as a "backend" storage mechanism, but since S3 files are available directly on the web, it would be nice to have the real filenames up there.
Is there a way to set that up? I know about publishing your files to the public, but it assumes you have a local git repo with all the data in the first place, something which may not be available...
my use case is that we have ~1TB of files already stored in S3 under specific filenames, and those filenames are how the files are accessed on the main website. changing all those filenames would be a significant burden... i'm not even sure this can be done cheaply on S3 in the first place. --anarcat
What real filenames? A git-annex special remote is a key/value store, where any key can correspond to zero or any number of filenames in the working tree.
What you're doing with S3 is using it to publish some files over the web. Fine. That's not a S3 special remote, so don't use it that way, just
git annex addurlthe files from S3.The new
git-annex exportfeature allows you to export a tree to a special remote, with the origiginal filenames being visible there.S3 should support it soon.