[tech] [spec] On extending gemini

[John Cowan]

On Tue, Feb 23, 2021 at 4:59 AM Nathan Galt <mailinglists at ngalt.com> wrote:


> Once some capsules start using metadata that?s only usable in some
> clients, those capsules will have an implicit Best Viewed In Gemini Client
> X label stuck to them. The Best Viewed In Browser X era of the Web is _not_
> something I want replicated in Gemini.
>

We are already there.

For example, a capsule that contains text/plain files can be rendered in
some clients but not all, or at least a conformant Gemini client can be
written that can't cope with text/plain.  The same applies to all other
media types, because media types are extensible.

What is more, a capsule that links to Gopher menus or documents is most
easily read using a client that supports either Gopher protocol natively or
an outboard Gopher proxy.  That's because of the extensibility of the
Internet protocol suite.

The issue that arose with "works best in browser X" was presentational:
some web pages looked good (for particular values of "good") in browser X
but not so good in browser Y.  That changed when browsers started to
include a JavaScript engine, and it became also a matter of what pages
would even *work* in browser Y at all.  But we are far from that precipice
now.

Fortunately, we have general agreement that there will be no *standardized*
presentational markup in text/gemini.  But it's still possible to write a
client that interprets a line beginning with "***" by rendering it in
italics, and we can't stop that, nor can we stop it from spreading to other
clients, or authors starting to use it.  That's because of the
extensibility of the interpretation of plain-ish text.

The only way to avoid all this is a WHATWG-ish standard in which everything
a client can do is spelled out in excruciating detail, and whatever is not
permitted is forbidden.  I think it is 100% unlikely that we will ever go
there.

Because I?m already strongly in favor of making it easy to make a good
> Gemini client, I?m against proposals that could, down the line, increase
> the amount of work that it?d take to make a good, full-featured Gemini
> client. Arbitrary-metadata proposals are a fully general can of worms that,
> say, double-digit status codes aren?t.


You can abuse metadata, but as I have demonstrated above, you can also
abuse existing content, and one is no worse than the other.  A metadata FAQ
would say that clients can convert the values of metadata lines from
machine-readable to human-readable (from "Jefferson, Thomas" to "Thomas
Jefferson", for example, or even from "Waldo, Edward Hamilton" to "Theodore
Sturgeon") but a metadata line should otherwise be treated like a text line.



John Cowan          http://vrici.lojban.org/~cowan        cowan at ccil.org
The present impossibility of giving a scientific explanation is no proof
that there is no scientific explanation. The unexplained is not to be
identified with the unexplainable, and the strange and extraordinary
nature of a fact is not a justification for attributing it to powers
above nature.  --The Catholic Encyclopedia, s.v. "telepathy" (1913)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailing-list-archive.gemi.dev/www/attachments/20210224/54e05a0a/attachment.htm>