This is interesting, I really like the idea of effectively loading the pages in reader mode in the first place, instead of needing to load the entire page first and then switch over. Using uMatrix...
This is interesting, I really like the idea of effectively loading the pages in reader mode in the first place, instead of needing to load the entire page first and then switch over. Using uMatrix to block scripts by default speeds up the web massively, and I think this approach should have similar gains for a lot of sites without being quite so complex and intimidating for people to manage.
I feel like applying machine-learning to this for figuring out which pages it can apply to is a little unnecessary, but whatever works.
If you start trying to build hand-crafted rules or heuristics about what parts of arbitrary HTML are essential, you’ll quickly understand why they went for a statistical classifier.
If you start trying to build hand-crafted rules or heuristics about what parts of arbitrary HTML are essential, you’ll quickly understand why they went for a statistical classifier.
I'm with you on the machine learning part. Seems like people are trying to fit that in wherever they can nowadays, but like you said, if it helps it's fine.
I'm with you on the machine learning part. Seems like people are trying to fit that in wherever they can nowadays, but like you said, if it helps it's fine.
I've often wished for a way for Firefox to just load everything in reader view, or be able to do a container tab that only loads things in that view. It would be great as a container because I...
I've often wished for a way for Firefox to just load everything in reader view, or be able to do a container tab that only loads things in that view. It would be great as a container because I obviously wouldn't want that view for say shopping on Amazon or similar.
This is super neat! I didn’t catch any details on how to use this, though. Can it be used outside of Brave? I have found the Mercury Web Parser to be servicable, but it’s not totally reliable and...
This is super neat! I didn’t catch any details on how to use this, though. Can it be used outside of Brave? I have found the Mercury Web Parser to be servicable, but it’s not totally reliable and means throwing traffick through their service. I’d much rather run something locally, esp. if it prevents making unnecessary requests. Like, I’d love if I could somehow use this with curl.
This is interesting, I really like the idea of effectively loading the pages in reader mode in the first place, instead of needing to load the entire page first and then switch over. Using uMatrix to block scripts by default speeds up the web massively, and I think this approach should have similar gains for a lot of sites without being quite so complex and intimidating for people to manage.
I feel like applying machine-learning to this for figuring out which pages it can apply to is a little unnecessary, but whatever works.
If you start trying to build hand-crafted rules or heuristics about what parts of arbitrary HTML are essential, you’ll quickly understand why they went for a statistical classifier.
I'm with you on the machine learning part. Seems like people are trying to fit that in wherever they can nowadays, but like you said, if it helps it's fine.
I've often wished for a way for Firefox to just load everything in reader view, or be able to do a container tab that only loads things in that view. It would be great as a container because I obviously wouldn't want that view for say shopping on Amazon or similar.
This is super neat! I didn’t catch any details on how to use this, though. Can it be used outside of Brave? I have found the Mercury Web Parser to be servicable, but it’s not totally reliable and means throwing traffick through their service. I’d much rather run something locally, esp. if it prevents making unnecessary requests. Like, I’d love if I could somehow use this with
curl
.I didn't see anything about it being used outside of Brave. You could ask them though.
It should be released by next year. They're not done with that feature yet.