A lot of discussion about column-fill:
stackoverflow
Clara Sexton "Since your CSS is almost 25,000 lines long..."

Richard Sexton I'm not sure this stuff is ready for prime time.

Barry Shein CSS is this interesting, even powerful, idea which went nuts in the details, particularly when one starts to deal with browser differences and lots of omitted details -- how do you style a range thumb in general? Maybe -webkit-/-moz-/-ms-, maybe not, maybe you use a trick to make it disappear and then write a new range thumb in javascript but then what was the point? But no standard CSS to address it) .
I suppose one could say that's not CSS' fault much like one might say that if we'd never standardized train track width it wouldn't be the trains' fault.

Richard Sexton In other words it's a bad add on to a bad language. And you need to add it on because the language is so useless in the first place.
And even then you have to add something more powerful than C just to make it work and oh there's an internal model (the DOM) that you'll need to learn all about too.
So, today in 2017 people need to learn 4 fairly horribly and complex languages (can't sleep, CSS selector syntax! Javascript '===') just to make words come out on paper pretty much exactly the same as runoff did in an old PDP 11/20 in Murray Hill in the 1970s. Hello? Oh, it's for you Sir Tim, Bell labs is calling, thanks you for adding color, but want their late 1960s imagine model back. Cause if there was any sense of what' right in the world then at least should have evolved to be more powerful and easier to love and not a machinated clusterfuck of a myriad of truly arcane and byzantine languages that have so little to do with each other you wonder how it's even possible to learn both with all their not so subtle diametrically opposed nuances.
After roff died a death by attrition and TeX failed to progress in the marketplace, HP filled that gap with their nutty and ever changing hpgl spec for the first laser printers and nothing else mattered for years.
Then postscript came out and threw all this stuff under a bus. It was clear it had won, the fonts worked for a change and it could draw anything because you sent a small program to the device not some godawful mashup of four different products form the remedial class in academic computing and god help you if you only know 3.
To this day you can master all of this crap and you stil can't draw an outlined letter. 'helvetica findfont 18 setfont hello world showpage' will do it, I think, I haven't looked at this in 20 years but that should be close. You can fake it in html/css+|js, but it's about 180 lines of code and only looks good in very large sizes. ps was notorious for making things look decent in small sizes. In the early 90s after a mutual friend show us html my reaction was this was a first cut and would be thrown away when something decent was ready. Nope, they were serious about this html garbage and it's staying. What special kind of idiot releases a worldwide publishing system that can't change fonts? Help.
The idea "why don't we use ps" was thrown out because ps was to hard to understand. Like anybody codes in html now, you can't it can't do anything.
7 postscript words vs 180 lines of incomprehensible garbage in three very different languages that doesn't work but is as close as you can get.
Keep telling me ps is to complicated. Seven. Words.
Browsers support pdf now. As soon as they support level 1 PS inline, this madness can end. We won't need addons for styling like css, that's already built in. We don't need js, lovely (but hideously broken) as it is because a language is already built in and we don't need html because we've outgrown roff thank. Hell even the guy that is responsible for the html Sir Tim lifted also said "use postscript". sglm/html was for document databases, ps to (finally!) achieves a rational sense of device independence and guess which one the web really is?
It's inevitable that browsers will do this, they're all consuming now and the sooner the better, the impact of .ps files over .html (+) files is a huge huge gain. The network will be faster than pages snappier, much.
Now, as to multi column, it doesn't work. It works in the limited case of their examples but as soon as you do anything complex it all falls apart, stops working and the only way out is a custom hybrid of flex box and.... fuck off.
And they wonder why people use tables? They work, that's why.
Never mind this was an 18 line program in the green book that always worked.

Barry Shein The whole web technology stack is a hideous botch. It's just IBM 3278 (70's/80s dumb terminal) gone mad, the 3279 even had color but it also had form fields and labels and you could even say a form field only took digits for example, you could specify a "picture" variable like 99999 which meant 1-5 digits in the controller and it'd beep at you if you hit some other key. And you could tab and backtab around. No real graphics tho so it wouldn't have been sufficient, it was for people who were entering your Time magazine renewal or airline reservation etc.
But look at games and real web apps and their interactivity and graphics going back to the 80s. X11 for example (and NeWS) were real client-server.
So instead we started all over again with basically a markup language, display this text in this font here, create one of five possible input fields here (text, checkbox, radio, etc.) A link, a box with a graphical image, maybe make the graphical image clickable, and so on.
And that's what we have, this form-filling mess with a lot of crazy add-ons which don't really do all that much but with great complexity and bad interactions. And a lot of incompatabilities but that's almost the least of its problems, the whole thing just adds up to form-filling on steroids.
Now look over the shoulder of someone playing a video game and imagine trying to do that in the web stack other than going full-bore javascript or java and just ignoring the rest. But no one really does that. Why? Because you can't control the complexity.

Richard Sexton The first thing in did in LA when I got there in 80 ways to make 8085 based systems fake out ATT's version of the 3270, called a "model 40". That was just video tab stops. html is *slightly* more sophisticated. But not by much.

Barry Shein I'm still convinced the reason Tim Berners-Lee wrote the whole web thing at CERN was to share MS help files. Back then 100mb (mega) disks were big so a place like CERN probably had over 1,000 copies of the MS help files spinning on people's desks. Probably 15MB of help files per each, imagine the waste. So why not have one or a few copies and just a server/client to fetch and display them? I know because I had the same idea at Boston University around the same time. Except I wrote xman (do a 'man xman' my name is still on there under AUTHORS) because I was unix-oriented and it was much simpler just run it thru troff and interpret it for X11 (actually X10 at the time.) And guess what? All those MS help pages were written in RTF which looks a lot like a simplified TeX but hey GML/SGML was on the table already and much simpler, you didn't really need a sophisticated typesetting language just bold and indent etc, so hey HTML. And voila, The Web.


Richard Sexton You might be right, there's a popular notion is was for "physics papers at CERN" but I think the exact phrase was "documentation to CERN". Which leads to the idea roff was invented to print manpages, the help files for unix, while the web was invented to make it easier for clippy to help you store ms help files. http://cdn3.meme.am/.../clippy-it-looks-like-you-used-15...