Michał "rysiek" Woźniak · 🇺🇦Feb 20

Web developers: we are going to go out of our way to hide the scrollbars!

Users: but that makes it difficult to know how much of the text is left to read!

Web developers: don't worry, we'll add a non-standard UI element that signals this in some non-obvious, weird way!

Show threadCassandrichFeb 20
@rysiek Spicy take: "Web developers" should not be a thing. You don't have any reason to "develop" for the web. The whole point was to factor things so that the browser and the server software were the only things that needed "development" and this development is reused by everyone. The actual documents you're publishing need writing/markup, not "development".
Show threadRotkäppchen
@dalias @rysiek
Well, there are legitimate reasons to develop applications running in the browser engine (portability as the main one).
The problem is article sites pushing advertisement bullshit into everything, and you can‘t really do that if the user just enables Reader Mode. So, make your document tree useless for everything but proprietary CSS and JS.
A web dev should just be someone developing applications for the browser engines, like iOS devs for iPhone apps.
Feb 20 at 3:31pmWeb
Show threadMarc 🐷🌏Feb 20

@libewa @dalias @rysiek
Imho the web should have stayed a gateway to information, not for serving cloud apps.

#localfirst

Show threadRotkäppchenFeb 20
@marc_eu @dalias @rysiek
But the janky web apps some companies write are at least sandboxed. To put it in the words of a friend „I‘d rather trust Mozilla than my bank to write safe code.“
Show threadCassandrichFeb 20
@libewa @rysiek The problem there is that vast, vast majority of "applications" on the web never should have been applications.
Show threadRotkäppchenFeb 20
@dalias @rysiek
Yes. But, to take one of my projects: a live departure viewer. Not enough for a native app, too much for a one-time GET.
Show threadCassandrichFeb 20

@libewa @rysiek Yeah but that's a document that can be complete in the moment as served, with some minimal js to make it an active document that updates elements of the DOM.

It does not need to be an "application" that bypasses the whole document model and implements its own UI.

Show threadRotkäppchenFeb 20
@dalias @rysiek
Ok, yes. *That* is despicable.
Show threadCassandrichFeb 20

@libewa @rysiek Prior to React, even stuff like Facebook followed the document model, where the document is semantic and the AJAX stuff just updates the semantic document.

React is what ruined the web with the despicable ideological position that the primacy of the document model is a bad thing and that you're supposed to write everything as an "application" that "renders" to the DOM.

Show threadSpark Purcell (they/them)Feb 20
@dalias @libewa @rysiek as always, the adtech industry is to blame. (the reason the "modern" web sucks)
Show threadMichał "rysiek" Woźniak · 🇺🇦Feb 21

@ailurocrat @dalias @libewa partially, yes. But there were scores of web developers that should have known better, and yet helped implement this.

This is also on them.

Show threadLisPiFeb 21
@dalias @libewa @rysiek It can even be a static HTML document in an <iframe> with a user-customizable (with a request with a form field) refresh timer enabled by <meta> tags.
<meta> http-equiv attribute - HTML | MDN

The http-equiv attribute of the <meta> element allows you to provide processing instructions for the browser as if the response that returned the document included certain HTTP headers. The metadata is document-level metadata that applies to the whole page.

MDN Web Docs
Show threadLisPiFeb 21
@libewa @dalias @rysiek > Not enough for a native app

Why not? I've seen similarly-sized "Apps for GNOME", they love small programs for every little thing.
Show threadLisPiFeb 21

@libewa @dalias @rysiek Java and Common Lisp are two better examples of how to solve the portability problem.

(Portable bytecode and runs-from-source, respectively, for native programs. Of course Common Lisp (and Interlisp) also did bytecode before Java but nevermind that, bytecode is old.)