I strongly agree with you... if you were to experience Windows 2000 Server with either Visual Basic 6, or Borland Delphi, on modern hardware, you'd be amazed.
Both had context sensitive help, which included example code. Both compiled to an executable that you could easily distribute, and would reliable run in most places.
Now we try to project a user interface through the internet, randomly chosen web browsers with ad-blockers and all sorts of add-ons, on screens from 8K or more, down to cell phones.
Help largely consists of the names and descriptions of parameters of the various libraries, and there are no working examples.
The frameworks get replaced at breakneck speed, and nothing is compiled and thus stable. Having source code is great for development, but over time all of the dependencies break the projects that use them.
Sure. And when something like Ramda comes along to clean up part of the mess, it's all negativity around here on how hard it is, how weird the code becomes, how team-work is no longer possible (posted two days ago on hn)
Meanwhile, chatgpt steals the show and minds of sw devs.
Oh, well. We'll probably get along with "terrible" for the foreseable future.
> We'll probably get along with "terrible" for the foreseable future.
They were saying much the same in the 1960s.
Chatgpt may be a blessing in disguise; when GOFSW (Good Old Fashioned Software) goes out of fashion there more be more interest in doing away with its bell bottoms and tailfins.
It's all a corollary of "good enough". It may cost sanity or rigor or beauty or clarity, but you can Make It Work anyway.
The entire modern web application architecture and development stack is a giant testament to that.
After all it is frankensteined to what was intended as a distribution architecture for hyperlinked, styled, mildly interactive documents.
W.r.t. web apps: Imagine there was a concerted, first principles effort to establish an architectural standard for cross-platform, network-insta-deployed, sandboxed applications. It may even reuse existing parts of web app tech, but apply them with clearer goal and scope.
Then there's questions of computational freedom and control...
Imagine there was no standard for electrical appliance outlets, no circuit breakers, etc, so every toaster wasn't automatically protected, etc.... That's the situation as I see it.
We need Operating Systems that default to no access, and only explicitly make resources (like power in the above example) available when the user decides to do so.
There is no central coordination in sw dev, solution rise and develop on a close-to-anarchy situation, maybe partially this is a global consequence of a market-driven development in large scale solutions (first come first grab best advantage), and standards develop after this.. a madness.
But there is a real alternative? Maybe a wise use of "old" solutions, lang, libs, standards, before reinvent the wheel again, but never quite round..
7 comments
[ 2.8 ms ] story [ 29.3 ms ] threadBoth had context sensitive help, which included example code. Both compiled to an executable that you could easily distribute, and would reliable run in most places.
Now we try to project a user interface through the internet, randomly chosen web browsers with ad-blockers and all sorts of add-ons, on screens from 8K or more, down to cell phones.
Help largely consists of the names and descriptions of parameters of the various libraries, and there are no working examples.
The frameworks get replaced at breakneck speed, and nothing is compiled and thus stable. Having source code is great for development, but over time all of the dependencies break the projects that use them.
Meanwhile, chatgpt steals the show and minds of sw devs.
Oh, well. We'll probably get along with "terrible" for the foreseable future.
They were saying much the same in the 1960s.
Chatgpt may be a blessing in disguise; when GOFSW (Good Old Fashioned Software) goes out of fashion there more be more interest in doing away with its bell bottoms and tailfins.
The entire modern web application architecture and development stack is a giant testament to that.
After all it is frankensteined to what was intended as a distribution architecture for hyperlinked, styled, mildly interactive documents.
W.r.t. web apps: Imagine there was a concerted, first principles effort to establish an architectural standard for cross-platform, network-insta-deployed, sandboxed applications. It may even reuse existing parts of web app tech, but apply them with clearer goal and scope.
Then there's questions of computational freedom and control...
We need Operating Systems that default to no access, and only explicitly make resources (like power in the above example) available when the user decides to do so.
There is no central coordination in sw dev, solution rise and develop on a close-to-anarchy situation, maybe partially this is a global consequence of a market-driven development in large scale solutions (first come first grab best advantage), and standards develop after this.. a madness.
But there is a real alternative? Maybe a wise use of "old" solutions, lang, libs, standards, before reinvent the wheel again, but never quite round..