I know #GeminiProtocol or #SpartanProtocol all are far from perfect (and will be hard to fix as they have no version by design to prevent additions). Still I believe they are the things that has taken off most and really means something symbolically about a separate non-commercial web.
Maybe it’ll all be rewritten with totally different protocols and content converted (it’s mostly markdown-like) but it’s still good to use & promote the version we have now.
Even though at this point #IneverDoAnything one of the aims of my #atomique project was to include a minimal client for smolweb. It would include basic Gemini reading and caching, as well as consulting markdown notes in folders.
It would be just C++ and run on ESP32 device like the M5 Paper but be portable to more powerful devices for a similar minimalist experience.
Considered using SyncThing possibly for #atomique data to make sync simpler (currently uses git). Installed on several machines on MacOS , Linux, BSD … before realizing it’s written in Go.
One nice feature of #atomique is that I can just bookmark posts here to add the URL to my read-it-later or watch-it-later list. One tap.
Even better: since I haven’t worked much this feature isn’t ready at all so I don’t actually read them. So it works exactly like Instapaper in practice !
In my purely theoretical world, I write about while lying down uselessly all these things are just simple text objects in my #atomique folder. Whether they are a task, an URL to read or watch. All of which could be interacted with in terminal or with a native client.
Downside: nothing exists yet because #INeverDoAnything . So far some bash scripts and musings about format definition. Still on the verge as to which language I should use for the native terminal client.
After reorganizing stuff I was left with a Pi3 with no life goal (just like me!) right on the day NetBSD 10 was released.
Was a bit lost at start with an unknown OS but overall it runs pretty well and has a lower memory usage than Debian. I only tried text mode stuff so far. Ok a younger me might have laughed at 116MB ram doing pretty much nothing in text mode being considered “low”.
@Sburlot Actually I now use two Pi4 on side screens that just run cool-retro-term full screen running my #atomique project (a work-in-progress which in practice today is just 90% fzf selection of SSH servers).
Thanks to synergy1 they feel mostly like part of my Mac desktop at the center.
Except not as smooth, especially when rotated on a vertical screen (Wayland is faster but incompatible with synergy).
The bookmark module in #atomique should not only have tags like good old del.icio.us but also periodic checking so they can be marked as down temporarily or if domain name does not exist anymore
In #atomique readlater module I want an option that auto scrolls downs slowly. Native iOS app could even check for attention - if eyes are looking at screen to stop scrolling.
Should also work to read mastodon lists.
Yes I want this to be able to read on border of pool while exercising with my arms :-) #decadence ?
Today my only real life use for #atomique is connecting to ssh bookmarks with fzf or as a launcher for other console apps.
The rest is just fiddling around but I keep imagining as a center for nearly everything in my online life. Like the only tool to publish, read, watch stuff. All with same minimalist text/markdown data structure described here.
Obviously nothing of this makes sense until the tools are actually written but the idea of it working uniformly for all this is exciting to me.
I don’t know if it’s a good thing but :mastodon: has become really central to most the stuff I read / watch and plan to do. Once I finish integration with #atomique I expect it to become even more. Right now the content I gather here goes to my mail, my todo list , instapaper all in a sort of unorganized way.
It’s a bubble for sure but a comfy one and one where I just discover a lot of stuff and simply waste much less time getting uselessly angry at political or other stuff.
Thinking about a minimalist / human editable pure text (not JSON) way of organizing data in #atomique. Store one file per item (notes, bookmarks, ssh servers, read later, watch later).
Attributes are just key:value. Attachments use a < separator.
File names are 4+3 lowercase alphanum.
Example: n000.not
title : Rice Pudding
url : https://some.link
created : 1676025600
updated : 1676025999
body < 8cd9.md
image < 8cda.jpg
tag : recipes
tag : indian
26 letters + 10 alphanum on four chars means 1.6 million files in each category. Enough for anything “human” in size. Should work also on most retro systems.
Not saying I’ll make an Amiga, Apple II or DOS client soon but I like to keep it possible.
Encoding is a big question of course . I’d go for UTF-8 for simplicity on anything remotely modern. This means any retro system accessing the data may need conversion / or accept loss when syncing.
Many retro systems didn’t even have folders anyway so instead of syncing I could accept an hypothetical C64 client would store nothing and accessed by serial (data on a some posix host). Only reasonably modern stuff would actually fully sync.
Prototyping is in shell scripts as it’s a nice way to think about limitations. If the bash commands to access something is too complex then maybe the structure isn’t simple enough. I want to keep it simple enough for shell scripts to be a valid platform.
Only thing I really use are SSH servers. Old format was one per line as I searched them with fzf. Switching to 1 file per item won’t be faster but difference for < 50 entries on a modern file system won’t be noticeable.
Opinion: I don’t think markdown / html should ever had offered more than two heading levels. One for chapter/article titles and one for sections inside the text body. You know, like most books. Having to melt lead for a new font size made you think twice about purpose.
When you give people six header level options they seem to get confused.
Maybe this could be integrated into #atomique once it becomes something native. It is currently only a set of bash scripts in the terminal so the typography is just fine :-)
One thing an ideal mastodon client could have : adding tags / types to bookmarks.
To me they can mean:
read later
watch later
todo item
I could to this in #atomique afterwards . Considering all bookmarks unsorted by default and marked for review in a GTD sort of way. Then only visualize them in respective categories after they get a tag.
There must be previous #INeverDoAnything posts summing it up but the idea of #atomique is to reduce file data to some very simple text based form so that clients can be done for anything. Syncing could use different layers depending on machine capacity. It could be a modern desktop but also say an Amiga or AppleII getting just the pre-parsed data via serial. It will always have a centralized small Linux machine doing the preparatory steps (currently a :rpi: as it’s only a shell prototype today).
Is there a good Peertube app for iOS ? I just want to keep a centralized “watch later” list. So just get the URL on any instance keep it with a thumbnail.
Actually what I want is an app that just gets my mastodon bookmarks to either Read Later or Watch later.
Of course I could write it myself. But #IneverDoAnything . I actually already did this in my #atomique shell prototype, just for the console (for reading texts only obviously, not ascii video :-) .
Now using Frogmouth , a #Textual app, as markdown reader instead of mdcat in #atomique . It looks cleaner.
It can get a bit slow on a :pi: with the initial parsing before displaying a large file. The clean up part could be instantaneous if caching wasn’t disabled.
Fiddled with #atomique to filter content for reading. All links in HTML are replaced by [1] or [2] and so on and moved to the end of the resulting markdown.
Viewing is done in mdcat at the end. Hopefully this will increase concentration reading.