zlacker

>>mig4ng+(OP)
This is cool. I did something similar at one point. Unfortunately these websites change their basic layout so often that it felt like these fixes would work for 1 month max then I'd have to configure again.

I ended up moving to news feed eradicator. https://chrome.google.com/webstore/detail/news-feed-eradicat...

I let myself use reddit for 5 minutes every morning. It auto blocks the feed when those 5 minutes are up. Every other site I just leave blocked.

>>Always+Ri
> "Unfortunately these websites change their basic layout so often that it felt like these fixes would work for 1 month max then I'd have to configure again."

You're exaggerating. My userContent.css is 60kB, and although breakages do happen indeed, it's occasional and nowhere near "redo everything every month".

What I will reckon is a pain, are machine-mangled CSS classes (e.g. by packers for React / other frameworks). They are kinda stable, until they're not, and at any rate, their inscrutability makes maintenance more difficult (because .user-profile-picture is human-transparent, while .cD5aZf is not :-/ ).

>>ronjou+Ir
Judging by some of the other examples here I’m guessing this is a case of writing fragile rules that e.g. count n and m items into a tree and test unnecessarily for incidental classes. If all you do is accumulate the often senseless outputs of the element picker I’d expect the gains to be short-lived.

>>aendru+mz
Indeed, 100%. I learned to never do such things, and would rather not have a rule than have a brittle one. This explains that my experience differs from OP's.

>>ronjou+jI
There are better ways to do it. For example, you can match elements containing specific text, such as the text that introduced the annoyance. On NewEgg you might match the text “Download our app” and use it to remove that whole box.

>>db48x+zN
Yup yup.

- Standard CSS (for userContent.css): https://developer.mozilla.org/en-US/docs/Web/CSS/Attribute_s...

- uBlock Origin: https://github.com/gorhill/uBlock/wiki/Static-filter-syntax & https://github.com/gorhill/uBlock/wiki/Procedural-cosmetic-f...

Still, sometimes it's difficult/impossible to make a reliable filter, and in such cases I'd rather not have it than have a brittle one.

>>ronjou+LO
Yes, I feel the same. Even though I still have some brittle rules, I avoid most and prefer filtering for specific text or classes/ids and find the correct element from there. Before I did not use uBlock to filter elements that much because I used the filter.

I wonder if you have your filter rules open source or available somewhere. Please share.

>>mig4ng+Zd1
I didn't make the effort to share them, sorry. That being said,

1. My rules are not any different from yours. I doubt you'll learn much from them given what you're already doing in your repo.

2. I feel that the "what to hide" / "not to hide" choice is too personal to be reusable by anyone else. I'm sure some of the stuff I hide will be considered excessive, and some will be considered missing. What I enjoy in this HN thread is that we share the { practice, tools, docs, tips }, then to each their own :)

3. I'm not interested in maintaining a public repo of that kind of stuff, and/or replying to Issues. So, would rather not make it public.

Sorry/notsorry ¯\_(ツ)_/¯ , at any rate, glad we're sharing tips around the practice.