Specify media types for the different CL dialects

[clange]

CL ontologies will be published increasingly on the Web. When publishing stuff on the Web, particularly when publishing ontologies in a linked data compliant way, it is good practice that the web server indicates, in the HTTP header, a media type for anything it makes available for download. (I can provide further reasons for this if desired.)

CL doesn't specify media types for its dialects so far.

For comparison, consider, e.g., OWL, e.g. http://www.w3.org/TR/owl2-xml-serialization/#Appendix:_Internet_Media_Type.2C_File_Extension.2C_and_Macintosh_File_Type

The different serializations of OWL have the following media types:

• OWL XML: application/owl+xml
• Manchester: text/owl-manchester Additionally OWL can be serialized as RDF graphs, reusing the RDF media types.

I suggest we come up with something like

• CLIF: text/clif
• XCL: application/xcl+xml

[greenTara]

What is the procedure for registering media types?

PS. Do you consider this a bug or enhancement?

[clange]

First of all, I'd call this an enhancement, as it would add something that hasn't been there before. OTOH, if bugs get higher priority, we could call it a bug, as a standardized data format should really have a media type defined.

Here is how to register one:

• General procedure: Specification/Registration (to supersede RFC 4288), RFC 4289, How the W3C does it (in their Last Call/Candidate Recommendation stage)
• Specific considerations for XML media types (applies to XCL): RFC 3023
• Specific considerations for text encodings (applies to CLIF): RFC 6657

In a nutshell I think it works as follows (I haven't read these RFCs in full detail yet):

• This is the registration form, but note that the W3C does things differently, namely by embedding the media type registration into their standards (cf. OWL 2 XML), which I consider good practice for CL as well. The RFC 4288 successor says that the media type specification (not the full CL spec!) needs to be published as a proper RFC, but I think that the W3C-specific procedure should also apply in our case. As the CL standard exists already, and I think our planned revisions won't affect the media types, we can probably apply as soon as we have that "media type" section for CL (unless we choose to rename CLIF or XCL – and of course we need to think about CGIF as well).
• CL being an ISO standard, our media types qualify for the standards tree.
• We must do an analysis of security issues, i.e. what harm downloading and processing ("executing") a CL text from the Web might do to a client.
• Additional considerations apply for text and XML media types (linked above). E.g. concerning CLIF, it is good that we already recommend using the UTF-8 character set, as IETF also recommend this for modern text media types.

[greenTara]

I have started to fill out the form at https://github.com/gruninger/Common-Logic/wiki/Media-Types

[greenTara]

For security considerations, so far I have the following:

1. spoofing of IRIs used to import other texts
2. use of untrusted texts in applications that have effects, such as decision support or logic programming
3. because CL is incomplete, the use of theorem provers on CL texts has the potential to consume large amounts of memory and processing resources, and to generate non-terminating processes.