4 ms·
I watched the video demo and I like the idea that I might easily get a PDF with nicer formatting than the default print styles. I would use that sometimes. Ho
by biztos 6y ago
I watched the video demo and I like the idea that I might easily get a PDF with nicer formatting than the default print styles. I would use that sometimes.
However, I question whether this will work consistently, since in my experience it takes a lot of developer attention to cover all the design anti-patterns out there.
It's a classic problem of course but I don't see why this is better than using something like Pocket, which has enough traction to get parsing right most of the time even if its print style might be less nice. It also has good integration with my e-reader.
For the intended physical-book version, I think you're setting yourself up for serious copyright problems. You're going to republish the Wall Street Journal because one of its readers asked you to?
Anyway it looks like a fun project and I wish you luck with it!
Regarding the video I'd like to point out:
1. It makes the service look really slow! Probably not what you want!
2. The video ends with an ad for the video hosting company, which I found confusing at first.
[edit: less speculation]
- yorgos_kats 6y agoThanks for this comment and taking the time, lot to unpack. From a technical standpoint, you are correct that this is a massive challenge. However for us to eventually get something print production ready, we needed to build our own extraction software that preserves stylesheet, images, sequence etc. We will definitely face a lot of failures initially, but hopefully will be able to build interesting machine learning models that will make the service much better over time. We have already built tools that will enable us to fix most of the possible issues pre-production. About the legal concerns, we will be printing content, we know 100% user already has access to, for its personal use. Won't be any different than creating a pdf file (eg from file print) and going to a print shop with a usb to print it out (or stitch). Regarding the video, it is slow, but this is the best we have made it to snapshot html, send it to our servers and extract the relevant info. We aim to make it faster, but that's the reality and didn't want to show anything fake. Rendering could take even longer depending on size etc, obv tons of room for optimisation.
- boffinism 6y agoRe the video, I get being honest, but no one wants to watch a video that's just endless loading animations. Instead how about: Shot 1: an article Shot 2: "30 seconds later..." Shot 3: the finished product That way I can see the product in action, know how long it actually takes to work, and not waste 30s of my life unnecessarily
- yorgos_kats 6y agoFair point and thanks for the feedback and suggestion!