💾 Archived View for gemi.dev › gemini-mailing-list › 000736.gmi captured on 2023-11-04 at 13:04:32. Gemini links have been rewritten to link to archived content

View Raw

More Information

➡️ Next capture (2023-12-28)

-=-=-=-=-=-=-

[Spec] <META> in the response header is too vague

Solene Rapenne <solene (a) perso.pw>



Link to individual message.

Mansfield <mansfield (a) ondollo.com>

On Sun, Feb 21, 2021 at 3:05 PM Solene Rapenne <solene at perso.pw> wrote:

> From the specification in "3.1 Response headers" it seems there are no
> current limitation to what could be used in META:
>
> <STATUS><SPACE><META><CR><LF>
>
> > <META> is a UTF-8 encoded string of maximum length 1024 bytes,
> > whose meaning is <STATUS> dependent.
>
> There is currently only the content type in it and possibiliy the lang
> metadata added for text/gemini types.
>
> Maybe I missed something about it in the spec, but it's really vague,
>
> - Is it ok to add many data in the response and up to the client to do
> something with them?
> - Should clients be strict and error in case of unknown fields?
> - What is planned for? It has 1024 bytes and doesn't use much, it seems to
> be ready for something else "just in case".
>

 My implementations of server/client use the META field to communicate
between the server and the user (through the client) - and in many cases,
the value in META *is* limited and well defined by the current spec.

See the 10 and 11 status codes - the prompt that the user should be shown
is the value of the META field.

The 3x status codes - the META is the location the server is redirecting
the client to.

Etc. (there are other places that META has specific meaning - even if the
exact value is left up to the implementor , like on status 10 - as the spec
says, the meaning of META is <STATUS> dependent)

Does that help?

Link to individual message.

Sean Conner <sean (a) conman.org>

It was thus said that the Great Solene Rapenne once stated:
> From the specification in "3.1 Response headers" it seems there are no
> current limitation to what could be used in META:
> 
> <STATUS><SPACE><META><CR><LF> 
> 
> > <META> is a UTF-8 encoded string of maximum length 1024 bytes,
> > whose meaning is <STATUS> dependent.
> 
> There is currently only the content type in it and possibiliy the lang
> metadata added for text/gemini types.
> 
> Maybe I missed something about it in the spec, but it's really vague,
> 
> - Is it ok to add many data in the response and up to the client to do
>   something with them?

  I'm not sure if "okay" is the right answer, but the MIME type (per the
spec) is defined in RFC-2045 (not RFC-2046 as stated---it's a typo) and
there, a consumer should be able to deal with parameters it doesn't
understand (read: ignore).

> - Should clients be strict and error in case of unknown fields?

  No.  Too hard to maintain strictness when new MIME types and parameters
are added.

> - What is planned for? It has 1024 bytes and doesn't use much, it seems to
>   be ready for something else "just in case".

  Nothing but parameters.  

  -spc

Link to individual message.

---

Previous Thread: [SPEC] Backwards-compatible metadata in Gemini

Next Thread: URL and URI links