Hacker News Reader

Comment by andy_ppp | Hacker News Reader

I think GitHub being an SPA is largely fine, it has good reasons to be in many situations, what I object to is when a shopping site or, even worse, a content site requires JS to work. This seems terrible when serving HTML is a much better experience for most users.

View on HN

No, it's not fine, there's absolutely no need whatsoever for Github to be an SPA. What functionality do you think can't be done ... in HTML 4.0 for that matter much less HTML5? Sure, a good editor inside the browser but then let that be a standalone, that has nothing to do with the repository viewer.

(As an aside I hate the Jira UI for being the same.)

I agree but I'll take it a step further. I actually rather dislike the in-browser editor when I'm just browsing code. It messes with my standard keybindings and has weird focus issues. Give me plain, syntax highlighted text, with optional symbol referencing to make it easier to look around. If I get to the point that I want a real text editor, I'll just clone it

That's what I meant by standalone indeed.

GitHub displays text! And is representative of a trend of websites that really only display text requiring JavaScript to work.

Which is fine ... but something has gone wrong somewhere if we require a full blown programming language to display text on the internet. It is absurd. It is one thing for HTML with a bit of extra JS on the side, but to be unable to display text without scripting enabled is comedy.

And if nothing else this really makes searching and indexing harder. That isn't good for the average internet denizen.

i liked seeing README immediately. I don't care that much for files list, source code. But README is the essence of what I'm looking for usually

I dunno, I semi-frequently want to look for a handful of files beyond the readme - license for obvious reasons, Dockerfile often tells me a lot about how to actually compile a thing, a handful of language-specific files indicate ex. what compiler version it needs. (Though I agree it's a pretty hard 80/20 rule with the readme being most of the value, then a tiny handful being the next 80% of the remaining value)

I'm really struggling to see why a shopping site, which is inherently interactive, would be less reasonable that github, which displays text. I guess maybe if you're using it as an in-browser IDE, but I'm not convinced that that's how most people are using it.

> This seems terrible when serving HTML is a much better experience for most users.

Tracking.

Pure HTML doesn't allow much tracking.

Our sector convinced c-suite that we need all kind of data, and now they want it.

It would seem that GitHub is (was?) not doing that, though? Remember when they announced that they dropped the cookie banner because they didn't need it: https://github.blog/2020-12-17-no-cookie-for-you/

Tracking isn't always bad, it is often being used to make the product better and discover usability issues.

I didn't make a judgement on the fact that it's good or bad for the user in my previous post, albeit my opinion is mostly negative.

Very poorly collected data, analyzed by people without much understanding about ux/product or data, and without any sensibility or intelligence took the place of thorough user interviews.

An example: in an ecommerce I've worked, a much better image gallery was released to show products than the previous one. All data showed users interacted with it much more. But it also showed conversion going down. The old gallery nobody used was then picked again and conversion went up again.

Now, what really happened was that users liked the gallery, what they didn't like where the pictures of the items. Those made them re-think buying it.

Thus, a crappier version of the website was released again, and users ended up having a worse experience, all in the name of the better conversion.

Seriously, with all due respect, I worked enough on front end and data collection to know that the whole tracking is generally harmful towards the user, slows down websites considerably, violates their privacy, leads to a worse experience and only rarely shows anything meaningful that user interviews wouldn't got.

It's mostly snake oil for c-suite, data "analysts" and product people and marketing so they can pretend they are providing any value. They generally aren't.

A much better way to discover usability issues is to have an issue tracker.

It doesn’t? I’m confused I mean no one used pure HTML anymore everyone has at least some JavaScript GA is a JavaScript file that can be used to track people pretty reliably. Pure html nowadays is paradoxically harder to maintain though.

Reply on news.ycombinator.com