File data meta gzip protocols

[shadders]

Generic 'file' protocol based on Ryan X Charles' proposal.

Generic 'data' protocol

'meta' protocol for marking up other protocols with JSON meta data

'gzip' protocol for compressing data elements in any other data element based protocol.

[ryanxcharles]

"filaname" should be "filename"

[ryanxcharles]

thanks for the credit :) minor issue, but for consistency it's best to have a period after the X - "Ryan X. Charles" :)

[libitx]

Over the weekend I drafted my own simple file spec, unaware that I was duplicating @ryanxcharles own efforts. Having now read the specs in this branch, I've decided not to send my own PR, but will share my work here in case my approach offers any merit to the discussion of these specs:

https://gist.github.com/libitx/ad28a004be12e9f331f58821de60b039

My own comments/questions:

• I believe the file and data specs cross over in a slightly ambiguous way. A file is essentially data with a filename, and I wonder if a single spec would serve better here? Perhaps using the data spec as the base, and where a filename is present combine it with the meta spec?
• In ryans file spec, it becomes the consuming clients responsibility to work out the correct media type based on the file name and extension. I think it should be the tx authors responsibility to explicitly set the correct media type, not relying on file extensions.
• In the data spec, it does become the tx authors responsibility to set a media type, but I think requiring longer form IETF RFC6838 media types would leave less room for error than the shorthand strings.

[ryanxcharles]

the advantage of my approach is that the code to "work out" the meta data already exists:

https://www.npmjs.com/package/mime-types

filename and filedata are the standard way to transmit/record files on every system.

[tonesnotes]

Filenames are valuable as long as they serve as identifiers, but they're weak identifiers. File extensions are likewise weak proxies for mime types. If the transaction ID (or a bitcoin address) are the real identifier, why not punt the filename and extension to metadata status and initialize the media type from the extension if possible?

[ryanxcharles]

but filenames already exist and are ubiquitous. email attachments, download from google drive, dropbox, etc. the "file" proposal is simply about putting files on the blockchain as they are already standardized.

[tonesnotes]

Okay. Makes sense. Maybe I'm thinking too much about reaching common ground. I do like the narrowly defined, composably collaborative approach of these specs.

[shadders]

Closing this PR again in favour of updates discussed in telegram..

#8