I would also question why we are saving collapsed comments. I feel like this is something that doesn't need to be persisted forever. If you want this behavior why not use local storage instead?
Is this the expected behavior when you collapse or show a thread? I would suggest that it is not, as most minor interactions on a page are only stored locally.
How many people actually use this feature? In order get a benefit from this feature you have to be logged into hacker news on multiple devices, collapse a thread on one device, come back to that thread later on the second device and then continue reading from where you left off. Alternatively you can just scroll past the stuff you've already read, which has a minimal cost to the user.
The benefits seem negligible for the amount of developer time that likely went into making this feature.
Both vote count and age influence the order of comments, which means that the stuff you've already read may appear below stuff you haven't read yet.
If you use the feature at all, I'm not sure why you wouldn't want collapsed comments to be in sync across devices.
I do.
> In order get a benefit from this feature you have to be logged into hacker news on multiple devices, collapse a thread on one device, come back to that thread later on the second device and then continue reading from where you left off.
I actually do exactly this. I read on my phone/laptop and later come back to the same thread (when more people have commented) on my home computer, usually at the end of the day. Having threads of conversation that you are not interested in already collapsed on multiple devices is very useful.
> Alternatively you can just scroll past the stuff you've already read, which has a minimal cost to the user.
Scrolling and quickly scanning over multiple threads is more time consuming and becomes actually annoying if you just want to passively follow a thread. If you want to follow multiple threads under the same topic, you either have to remember unique words/usernames/dates of that thread or resort to scroll & scan method mentioned above. One is annoying, the other is simply non viable for N amount of threads.
Until there is a way to subscribe to a thread, this feature is super helpful. Its actually just as good or better than a subscribe mechanism because atleast you can avoid getting multiple notifications for every single comment that get posted in the interesting thread. In the current scheme, all the irrelevant ones are already out of your way.
I don't want to scroll over all the replies just to find the next post or i want to see all replies to one post. Therefore collapsing is convenient, but the slowth is annoying.
This is awesome! I had no idea! You're right, I don't think most people even realize. There's often times I get upset when using Bootstrap or just doing CSS and look up ways to do specific things and don't find out of the box solutions, this is one time I'm glad I found a decent alternative. One case that comes to mind is collapsible trees and dropdown boxes where you can type / search on them, and anything with dropdown boxes like checkmarks gets a wee bit hacky.
18% of my users are on IE11. Some industries (healthcare, for example) don't change if they don't perceive something to be broken.
source: did support for a doctor some years ago
This reminds me: What happened to the markup rewrite that was announced almost 5 years ago together with the Firebase API? https://blog.ycombinator.com/hacker-news-api/
As far as I can see the markup never changed after all.
It was a long time ago now so I can't remember much specifically any more, but I think that the styling and behavior for <details> were both too limited and too inconsistent across browsers.
I do use it in some other places though, like collapsing/expanding the text of text topics from the listing pages, and it works really well for that (other than the current lack of support in Edge).
Its also missing a group name, so I can expand only one details of a group.
Why is it useful?
My guess would be that the decision was made before a standardized, widely available cross-browser implementation of localstorage was a thing.
I can't be the only one who had no idea it did this? And if so, I'd could argue that there isn't really a need for it.
Now that everyone knows about it people will start using it and people will start noticing the slowdown.
Somethings are best not mentioned