pendell

Pendell

@pendell

Current Hyperfixation: Wizard of Oz

  • He/Him

I use outdated technology just for fun, listen to crappy music, and watch a lot of horror movies. Expect posts about These Things. I talk a lot.

Check tags like Star Trek Archive and Media Piracy to find things I share for others.

log inask
Bluesky
bsky.app/profile/pendell.bsky.social
Mastodon
mastodon.social/@pendell
Bearblog
pendell.bearblog.dev/
Letterboxd
letterboxd.com/aependell/
pendell
@pendell
micolithe
@micolithe
NireBryce
@NireBryce

maybe the gap people are trying to fill with AI is the impossibility of catching up in a world that's heavily specialized, but largely hasn't taken the effort to re-compress that complexity into routes that don't rely on you knowing the abstraction stack that gets you there

NireBryce
@NireBryce

I forget who it was, but an old 'blogposter once said that he didn't understand why people were lamenting that "the new generation of junior programmers ("first years on the job" kinda) didn't know anything at all about how the underlying hardware worked".

His rebuttal was that the previous generation succeeded so thoroughly that the new one no longer needed to learn the entire abstraction stack to get started, and instead will eventually use that head start to do the same for the next one. the older generation (we're talking about like, 30 year timelines not actual generations) will need to keep alive how things work, and there will need to be specialists who learn it, but it is no longer a prerequesite to do software dev at all.

and i wonder what magic would happen if we actually took that seriously, in computing in specific, but ideally in every industry and field

You must log in to comment.

in reply to @NireBryce's post:

NireBryce
@NireBryce

yeah but it even falls apart with discord. most of the issues there are business, not like, core CS concepts of the current era (and the more modern languages have already done some of this work, but only for their language, and never the same overlap)

login to reply
queerinmech
@queerinmech

that is kind of my point, the business issues are what drives technology adoption, and while there are certainly major issues with the Discord organization specifically, Electron is a fairly tall house of cards built on layers of corporate decisions

login to reply
NireBryce
@NireBryce

right, but imo that's a symptom of a failure (as yet) of collapsing the abstractions to something you don't need to know to build things without having to use electron to do it. it's plugging a hole because actually doing what electron does for features requires knowing many stacks, and because of that it's shortest path. not an argument against the concept, (accessability, font and UI, embedding content (if you aren't already doing js), audio io integration with low latency, internationalization, scaling, off the top of my head)

login to reply
queerinmech
@queerinmech

the featureset of something like Electron is not itself bad! it is just a variation on a theme like a game engine or any full fat UI framework like Delphi, these are all useful abstractions

but i feel like most modern frameworks are these lumber behemoths - necessitated by their corporate stakeholders (and related to my commentary on modular systems a while back), Electron is just the ur-example

there is nothing wrong with using a high level of abstraction to get things done

it is just that many of them come with severe downsides - oh there is a bug in Electron? good luck getting anyone to acknowledge the bug and even better luck trying to fix it - particularly for the people who only have experience with high levels of abstractions, they are trapped at that layer

for example in 2022 i interviewed someone who had like 10 years of experience and who rightfully felt like an expert in development, but during the interview it became clear they had been working entirely with some proprietary high level framework (i forget the name) at one company all that time and they had a huge knowledge and skill gap to transition to any other tech stack and had no understanding of how anything worked beneath that layer, it was heartbreaking to watch them realize it in real time

i do not blame them for not spending hours every week doing research and testing other languages and frameworks like i do

i have seen the same happen with any big framework, like Django or Rails, and when something goes wrong beneath that layer of abstraction i have seen entire teams flounder

the level of investment that companies put into their teams is criminally low and encourages these outcomes, the problem is endemic in the industry

Electron is just one particularly damning example of it

login to reply
NireBryce
@NireBryce

right -- but imo it's because we haven't managed to compress those separate things into things that don't take all that wandering to figure out, that we get things like electron filling that gap in the first place

like this isn't me fighting for the industry's way of doing it, and some of this is just the tyranny of time, but a lot is how large swaths of rank and file in the industry don't really see themselves as like, the idea of contributing to the field in ways other than making software that papers over it (not that many of them are given an option)

login to reply
queerinmech
@queerinmech

i agree, i think there is still a lot of room to continue to develop better abstractions and better ways of describing interfaces in particular with an underlying library that can support creative and stable development

this problem is something that occupies a big part of my mental cycles, but there is little appetite for it in practice because "good enough" and "established" are some of the most powerful features in human society

login to reply
-pegasus
@-pegasus

agree with you.

i don’t like hypotheticals. anything you want can be correct in a hypothetical scenario because you imagined it in your head. we have to face the realities on the ground: there is no good high level abstraction in real life today.

everything between win32 and an electron user’s work is a jenga tower of 20 years of “it wasn’t designed for this originally, but…”, “how do we do this as cheaply as possible?” and “i can’t go home if i want to meet deadline so i sleep under my desk and the compiler beeps when it’s done to wake me up”. producing good software on this foundation is impossible.

electron is very good if you would like to push a website designer into doing software development because they demand less salary. that is the number one driving force behind it. the output merely has to be tolerable.

login to reply
jcolag
@jcolag

Yeah, I get maybe-irrationally angry whenever I see the "kids these days" whines, or the "you don't need [popular library]" arguments, because you (self-important pundit explaining why your education was inherently better) definitely do not know how to design and fabricate your own microprocessor.

In software, people who don't know everything that we know are worthlessly ignorant, and the people who know things that we don't are worthless fossils. Convenient!

login to reply
Pinned Tags