Hacker News Reader

Comment by throw_m239339 | Hacker News Reader

I really hated this era before flexbox and co. CSS used to be the least intuitive thing on the planet yet a lot of people claimed otherwise. We had to resort to silly tricks (remember overflow hidden?) in order to make up for the mess that CSS used to be. Again what really pissed me off isn't CSS inherent issues but the fact that many people claimed there was absolutely nothing wrong with it, a pattern I often see with many developers about this or that language. Obviously, there were some problems or we wouldn't be needing CSS grids or flexbox today.

I don't understand this.

You can appreciate something while still acknowledging its obvious flaws, nothing is perfect.

View on HN

In my experience, no matter how many improvements there have been, there are always things that you are trying to do that css doesn't support or make easy to do. Recently I've tried making tabs the same width but not justified horizontally, meaning all tabs should have the same width as the widest one. Something like this (1) but without hard coding the children count. In my mind, you would be able to set the widths to max-width of siblings. In general, I would love to have more signifiers that describe relations between elements.

(1)https://codepen.io/brettdonald/pen/oRpORG

Then what direction of relationship would have precedence, left to right? Right to left? If neither how precedence would be defined? Overall I don't want css to become any more complex, if you need an edge case such as that one JavaScript is a great tool for the job.

Funny, I was just struggling with this yesterday in an assessment I was given for a new job.

Yeah I have the same feeling. I was struggling for hours the other week to make two columns the same height, but the need for an overflow on one kept on killing everything.

Coming from a mechanical engineering background, I'm very used to being able to define spatial properties with respect to other elements in CAD, and it just seems lacking in CSS.

I find it really frustrating when project runners refuse to acknowledge shortcomings. I'm not always expecting them to solve it, but I'd at least like an acknowledgement that the issue I'm facing is valid. Pretending it's just me or that the obviously worse alternative is okay gets annoying.

One example is ReasonML's async story:

    Q: What's BuckleScript's async story?

    A: If you're not interfacing with any library that uses promises, you can simply use callbacks. Everyone gets them and they're performant. If you need to bind to a JS library that uses promises, or communicate with such library, you can use BS's Js.Promise.

IMO that's a less than ideal response. I'd prefer a more honest response like:

A: Unfortunately, we do not offer an async/await equivalent currently. BuckleScript has a Js.Promise type for using Promises. We understand that this is not ideal, and are working on other options.

Look—it's okay that they don't have the feature. Nobody's perfect. But they shouldn't pretend that callbacks are an acceptable replacement because they're "performant".

This is my biggest bugbear (the general behaviour, not the specific example). I want to make good technical decisions, your library has made choices based on its principles and design goals, don't obscure the trade-offs you've been made -- this is just a (poor) growth hack in disguise.

I've seen so many people use technologies then regret it a year later because they realised what they actually wanted to achieve was at odds with the design goals of the technology.

OCaml has monadic let syntax[1] which is similar to F#'s computation expressions. I am not sure why it is not yet supported in Reasonml/Bucklescript.

[1] http://jobjo.github.io/2019/04/24/ocaml-has-some-new-shiny-s...

Because Bucklescript/ReScript is based on a modified fork of OCaml 4.06, before monadic let was added.

https://github.com/rescript-lang/ocaml "This branch is 387 commits ahead, 5444 commits behind ocaml:trunk"

Perfect use of "performant" as a nonsense pseudo-adjective, to signal sarcasm and condescension.

If they are semi-serious, then "cromulent" would be a better synonym for "(marginally) acceptable."

Yeah, it might have embiggened their presentation

I've noticed that some people mistake 'the best extant X' with 'X has no problems.' For instance, 'Y is a problem with X' will often receive the response 'Nuh-uh, because extant Z also has problem Y but is worse than X in other respects.' Look for this particularly in conversations about smartphones.

> > iphones have poor battery life, I'd like a phone that keeps a charge for a full week at least.

> No you're wrong because android phones also have bad battery life.

I've also noticed this. It seema to be especially common amongst proponents of capitalism:

>> 21st century capitalism a wealth concentration problem

> No you're wrong because communism is terrible.

It's tough to know that something is bad when you don't know about anything better. I remember when I started doing Android development was when I really realized how terrible CSS was. And this was long before all the new constraint stuff in iOS and Android that still makes CSS look bad.

i suspect you might be mixing up people who talked about the spec and actual end users - at least can't seem to recall anyone ever saying css is good as it is, back in the days. The shortcomings where well known and anmoying to anyone having to work with them. But that might be true to my bubble only.

Seems like a bit of a strawman, people who couldn't think of a single issue with CSS. You are generally talking to beginners when you encounter anyone like that on any subject. Kind of like trying to draw too many conclusions from a Twitter argument you had with someone who is actually a 14yo (probably happens a lot more than we like to think).

Instead I think I could agree with people who could point out that it's all trade-offs, and there are some upsides to having fewer autolayout options in UI programming. For example, flexbox and cssgrids enables things you simply wouldn't have attempted before in CSS (you'd have gone with something less expressive, like media-query hard-coding). And any time you nest grids and/or flexboxes you get to rediscover the horrors of autolayout flexibility.

UI is hard and there is no "best". Though it's the emergence of mobile devices that forced the hand to more complex autolayout tools.

Reply on news.ycombinator.com