💾 Archived View for nytpu.com › gemlog › 2024-07-10 captured on 2024-09-29 at 03:34:35. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2024-08-18)
-=-=-=-=-=-=-
There's a webapp I've had a vague idea about making for a long time, and earlier today I was thinking about not having posted on Gemini for a while; so I figured I might as well braindump about the idea and flesh it out some more.
Do I expect anyone to make anything even somewhat resembling what I describe here? Hell no! (there's already a dearth of even basic bookmarking utilities like Pinboard) Do I even expect anyone to read this post? No, it's mostly just serving to conceptualize how feasible it'd be to make myself (spoiler: it isn't). Although if you'd also be interested in something like this, I'd love if you let me know!
I guess I should take a moment to discuss my very unusual method of maintaining a “knowledge base”. For the life of me I can never maintain a personal wiki or zettelkasten or even just notes on things. I take notes when necessary, like in a class lecture; but if I'm just reading something there's no way I'll take notes, mostly because I simply don't need to. I genuinely just remember the general gist of things I've read and the fact that I have it archived, assuming I found it at all interesting. So instead, what works best for me is to just maintain an archive of links and saved documents that I can search, so I remember there was a specific <book/article/webpage/lecture I took notes on> that discussed <topic> and can then search the archive for the specific document to get all the details.
So basically I want a knowledge program that specifically accomodates my strange thing in the same way that something like Obsidian specifically accomodates people that zettelkasten works for.
There's a number of relatively independent things I want it to do, but then integrate all those resources into the single searchable knowledgebase:
Then, really all I want is full-text search of all of those resources (when full-text search is possible). I don't really care much about tagging or manually sorting/interlinking things except in very specific situations, and I especially don't like when it's de facto required in order to locate anything in the future. That adds way too much overhead to the system and gets in the way of the actual archiving.
But the big revolution would be having all this in one unified system, because currently if I'm a little hazy on what the specific resource was, I have to search Pinboard, then Wallabag (which has really janky search which often doesn't find something that should've matched), then my various PDF, epub, website archives, and digitized school notes in git-annex (irritating since full-text-searching a pile of PDFs and HTML using generalized tools is basically impossible, so I end up navigating by filenames/folder organization). git-annex also prevents me from adding new PDFs and such from my phone, and storing things in a plain old directory doesn't allow storing associated metadata like the original source unless I manually create a “sidecar” text file that I paste the link and description or anything else into.
Unfortunately, I'm likely never going to build something like this. Oh sure, I *could* make it, but the final result really wouldn't match my vision for what I want. But mostly I don't even want to try because anything touching the modern web demands running on an incessant maintenence treadmill as technology Improves™, which is basically the exact opposite of what all of my other personal projects are: after the initial implementing, they are then essentially complete and then require no ongoing maintenence for years afterward, even if I use it every single day. The best outcome would be using external libraries for the web stuff so my code itself could stay mostly static, but unfortunately all the requsite libraries are in a webshit-focused language I absolutely do not want to use to write a server-side-rendered app (e.g. readability.js). I'll tolerate JS for the bits of code that need to run in the client, but hell no for all the business logic on the server.
⁂
contact via email: alex [at] nytpu.com