50 comments

[ 3.7 ms ] story [ 116 ms ] thread
I'm not certain that an SVG file counts as a web browser.
I can't even see the SVG file. It's blank.
Scroll down or zoom out.
You have to scroll to the right
The upper left of the SVG is empty (blank or white) so if your viewport is relatively small then you will be looking at a white page.

Did you try scrolling? It's a comment thread rendered as a graph; the root comment is in the top center.

Interestingly, most of the threading WWW discussion fora, and threaded mail/news readers that I've encountered start such diagrams in the top left hand corner.
Sounds like an incredibly poor think to post to HN, even if it may be possible to find germane content somewhere in the SVG.
I had to zoom out to 25% of the original zoom level to see anything.
It's a big, sprawling conversation, and the SVG gives a way to see the extent of it before you click on any of the nodes to dive in to the bit you're interesting in.
You seem to be confusing this specific SVG presentation of the conversation with the content of the conversation.
It doesn't say what browser they're talking about

Also... what in the world is this thread format?

You can click on one of the messages in the SVG to be taken to the source message, which also shows a more usable threading view.
That's the global view. If you have the full toolkit then you can have a "Neighbourhood" navigation mode that lets you move around while keeping a global sense of where you are. That's still experimental.

If you prefer the usual linear thread mode, click on one of the nodes and it will take you to the toot in question.

This SVG is possibly the most insane format for a forum I've ever seen. What CMS produces this, and um...why?
It's funny to see someone bitching about the modern web, who thinks an SVG graph is the appropriate format to share a discussion in.
Why do you think that Colin Wright and Göktuğ Kayaalp are the same person?
I actually had a cool experience opening it on the mobile. It sucks on the desktop though.
Large, sprawling "threads" are always going to be a nightmare to navigate. This is the way I find it easiest.

If you don't like it click on any node ... that will take you to the usual linear presentation.

Reading thru the thread, it appears the browser they are referring to may be Cutebrowser.
Yep, took me a bit to figure that out. I'd never heard of it, sounds like a fun way to find broken websites though...
Ah yes, welcome to 1998 again. Terrible companies doing stupid things to filter out browsers instead of making browser companies do the right thing.
FWIW, it is likely not the websites looking at user agent.

Just checked with "qutebrowser (1.1.1-1ubuntu0.1)", and both github and gitlab work, at least for code viewing. Github shows a badge but then works anyway.

The OP does say things like:

> if you choose to do sensible things like blanket-ban 3rd-party cookies, disable referrer API, disable canvas, etc.

so maybe they disabled too much stuff in the otherwise-working browser?

The websites should work without that stuff. Chase did something recently where they blacklisted browsers, its just another case of "Whats old is new again" in terms of bad software development.
We don't know what "etc" includes. If they disabled JavaScript for example then I'm not at all surprised the site wouldn't work.
Disabling JS shouldn't break a website. I refer folks to 1998 again...
If you're not allowing all cookies and allowing all JavaScript, today's web is pretty much in shambles.

I browse just with cookies limited to sites I have a relationship with, and it's… not pretty. A fair number of websites fail to load, or fail to load pictures. Twitter is hit or miss.

I'm working on a Web Extension that pretends cookies are enabled, and just black-holes them, or holds them in RAM for the single load of the page.

I run NoScript and I basically only browse hn these days, literally. Half the links here don't work, either, so I usually don't even read those. Twitter is a wasteland for me. I've accepted it, and I enjoy being an antisocial technophobic software engineer. I'm even thinking about only using electronics for work. I'm sick of the internet today, I feel like it only tears people apart.
I use the Firefox "Temporary Containers" addon (and Multi-Account Containers for permanent things). New tabs open in a temporary container, a few minutes after each is closed they delete all cookies/local storage/history/etc. Basically does what you're looking for AFAICT.
Almost. Temporary containers works for that 'session' while they're talking about doing it per page load.
You can actually do it per page load. Under Isolation, Global, Navigation, change "target domain" to "Always" (for a new container every link you click, no matter the domain and subdomain). There are also other options, eg to not create new containers if link is in the same domain (or domain and subdomains) as the current tab.
The poster mentions using Canvas to fingerprint browsers (which is new to me). Is there any obvious way to block it on Firefox?

I found this addon (https://addons.mozilla.org/en-US/firefox/addon/canvasblocker...) but not sure if this is something that really helps or not.

Blocking canvas fingerprinting makes you stand out like a firework, in fingerprinting terms.
There is an about:config setting called privacy.resistFingerprinting that does a handful of things, I believe including canvas fingerprinting. It also disables site specific zoom though (as that could be used for fingerprinting) which I find rather annoying.
I don't think the poster is correct about this.

Among other reasons, there is no "profit" from fingerprinting the browser. Fingerprinting the user allows for collecting user data that can be sold. Fingerprinting the browser has no such benefits -- the browser checks are usually done so that site "looks good" for an average user, and the amount of users who use rare browsers _and_ spoof their user agents is too small to care about.

(comment deleted)
Just for clarity, I just installed qutebrowser and tried it on github.

The page works just fine. There is a small yellow badge on top talking about changes which disappears after a few page changes, and that's it. The quick test actually shows the functionality works -- I can browse repos, view files, PRs, get git commands for checkout and so on.

I don't think there is much shaming needed in this case.

(GitHub CEO)

It looked at first like this thread was complaining that GitHub doesn't work in something called Qutebrowser. So I downloaded Qutebrowser to try it out, and GitHub seems to work just fine.

We care quite a lot about web standards and broad browser support at GitHub, so if anyone is able to tell what the issue is, please let me know: nat@github.com. Thank you!

> We care quite a lot about web standards and broad browser support at GitHub

From [0]:

> We design GitHub to support the latest web browsers. We support the current versions of Chrome, Firefox, Safari, and Microsoft Edge.

Following "web standards and broad browser support" and supporting recent versions of the four highest-budget browsers (three engines) seem different. There are at least a hundred browsers out there; keeping websites simple and standards compliant is a much easier route than testing on hundreds of different configurations.

On a semi-related note, to clear up any confusion: Qutebrowser uses QtWebEngine, which (like MS Edge) is Chromium-based.

[0]: https://docs.github.com/en/github/getting-started-with-githu...

I've seen this message in my RSS reader on my iPad.
As a daily qutebrowser[1] user, I am not sure why OP is facing these issues. I have the browser configured pretty tightly[2], with a bunch of things disabled—including JavaScript and canvas reading—and GitHub works well enough for my usage.

I am all for bashing GitHub and try to stay as far away from it as possible[3], but I do not think this one is on them.

[1] As in, qutebrowser is my browser of choice and I use it every day for personal and work.

[2] https://git.sr.ht/~jamesponddotco/dotfiles/tree/master/item/...

[3] I also much prefer the SourceHut workflow.

Why are tools like github and gitlab like this? All the work has to be done server side, the JavaScript stuff (especially hiding the tree when the window is smaller, that's the whole point of the thing!) is just annoying.

Github's PR UI is nice but it's times like these where I wonder if it really improves on a mailing list and a few cgit instances.