Steven Black DNS blocklist blocked gstatic.com | Hacker News Reader

Steven Black DNS blocklist blocked gstatic.com

github.com

5 pointsLeoPanthera7 comments

Open articleView on HN

While it is now unblocked, the Steven Black list has been blocking a lot of innocent CDNs.

jQuery: https://github.com/StevenBlack/hosts/issues/2520

(twice): https://github.com/StevenBlack/hosts/issues/2516

cookielaw: https://github.com/StevenBlack/hosts/issues/2519

jsdelivr.net: https://github.com/StevenBlack/hosts/issues/2518

gstatic (again): https://github.com/StevenBlack/hosts/issues/2329

And other innocent sites:

freecodecamp: https://github.com/StevenBlack/hosts/issues/2517

Parts of the Internet Archive: https://github.com/StevenBlack/hosts/issues/2476

I agree that these may break a lot of functionality, but I am pretty sure that these are not "innocent" in the sense that they wouldn't do tracking.

That's like saying we should block port 443 since it's not "innocent" because tracking happens over it.

No it is not: distributing small-scale JavaScript loads to third-party CDNs cannot be compared to blocking a port. The thing is, you could easily enough host those locally. There are also GDPR violations here, and there are plenty of security threats involved too. (Note also that it was the parent poster who used the term "innocent".)

Ive always "blocked" gstatic.com. Also ssl.gstatic.com. It has no practical use and does not "break" anything for me. The software I use has no need to access it.

A question: I read a comment of yours from last year concerning your bespoke browser. Do you transform websites that have a large amount of irrelevant formatting like sidebars?

Would need to see the comment. If I'm using a browser, I use a few different versions of links v1 or 2 with some relatively minor customisations.

As opposed to "surfing", I prefer to make HTTP requests outside the browser and use custom filters to manipulate retrieved text. Generally I'm only looking at HTML in links, offline, using the browser as an HTML reader. It's text-only, no graphics.

If provide a sample webpage maybe I can demonstrate.

Reply on news.ycombinator.com