Post AIUCSk5eZqoGUtNFeS by josias@floss.social
(DIR) More posts by josias@floss.social
(DIR) Post #AIUCSk5eZqoGUtNFeS by josias@floss.social
2022-04-15T21:38:20Z
0 likes, 4 repeats
I'm working on an alternative media type for #Gemini. I expect it to be contentious, so I'd like to hear some thoughts on what y'all would like to see in an extension Gemtext.
(DIR) Post #AIVSWn0TMQvipuQaq8 by devinprater@devin.masto.host
2022-04-16T00:06:52Z
1 likes, 1 repeats
@josias A way to tell the browser the difference between a code block and an ascii graphics block so the browser can just block ascii junk.
(DIR) Post #AIYTAFEavYZt9Qj86K by vidak@zoinks.one
2022-04-16T08:45:53.507644Z
0 likes, 0 repeats
@josias display images?
(DIR) Post #AIYTAFvUM4tBISnNxY by josias@floss.social
2022-04-16T13:52:22Z
0 likes, 0 repeats
@vidak Gemini clients are only supposed to make one request per page, so embedded images wouldn't work in any spec-following client.I think Lagrange's approach to inline images in existing Gemtext in as good as we'll get, allowing you to click an image link to load it. It already works pretty well.
(DIR) Post #AIYTAGe9g0cNWzh3a4 by Seirdy@pleroma.envs.net
2022-04-17T23:07:31.942253Z
0 likes, 0 repeats
@josias @vidak Yes, this is a feature. One request per interaction eliminates most fingerprinting and bandwidth concerns. Click-to-load is actually such a good idea that make all media elements above 100kb click-to-load in uBlock Origin.The only Gemtext feature I want: a distinction between pre-formatted text and ASCII-art, and for programming-language-indicators to be explicitly made a non-use-case for alt-text OR for code snippets to receive their own defined semantics. Really, really important from an accessibility standpoint (esp for screen reader users). devinprater@devin.masto.host aired similar concernsI don't think adding footnotes are necessary, since you can just add a paragraph and start it with the phrase "sidenote: ". Users can skip the paragraph if they want.I don't think adding inline formatting is necessary or desirable, since many clients (including many screen readers) just strip inline formatting on the Web anyway. Authors should try to express meaning through words instead.One of the best things about gemtext is links getting their own lines. This makes authors more likely to have more descriptive links that can double as navigational aids. Enhanced link meaning and the navigational aid make screen reader life much better.Some non-gemtext Gemini features I want:1. Compression. Technically not a Gemtext feature. Users of low-bandwidth connections benefit from being able to download single large pages instead of many small resources (see the "Against lazy loading" section of https://seirdy.one/2020/11/23/website-best-practices.html#against-lazy-loading). Combining many articles into a single resource can help with this, but payload sizes can skyrocket easily. Text compression can bring payload sizes down considerably; it's what allows many people to offer full-text RSS Web feeds that are hundreds of thousands of words long.2. Something to supplmeent TOFU so that the initial request can't be intercepted and altered. Since Gemini already relies on DNS, perhaps DANE could work. Alternatively, some sort of distributed hash table (no blockchain obviously) could help. Honestly, your best bet is to make you capsule available as a Tor hidden service (known on the Gemini space as "Deep Space capsules").
(DIR) Post #AIZzZuZ5n0qYGaYbRp by josias@floss.social
2022-04-18T16:32:13Z
0 likes, 1 repeats
@Seirdy @vidak In my opinion, Click-to-load is a great compromise between inline images and external ones, and actually fits fairly well into Gemini.1. As you mention elsewhere, wouldn't compression be a server and client feature? Gemini doesn't forbid sending or decompressing gziped files.2. As for TOFU, I liked the concept of also loading the capsule via Tor alongside the first request to see if the certificates match before adding it to the list of trusted certificates. Could that work?
(DIR) Post #AIa0BuuUxDraLEgxuK by Seirdy@pleroma.envs.net
2022-04-18T16:52:19.533656Z
0 likes, 0 repeats
@josias @vidak For no. 2, this tells you that your client's connection is not compromised. It does not tell you that the server's connection is not compromised.