~Originally posted by Chaser
Requested feature overview description.
Dynamic pagination. I was writing code for my private local booru* and though E621 could benefit from this as well(speed wise).
Basically: On load of /posts/*, instead of sending the entire list of images to the browser, request /posts/*.json instead, applying the blacklist when creating each entry for the thumbnails.
This also allows for infinite scrolling.
Also, only load the thumbnail once the browser can actually see the thumbnail image.
Why would it be useful?
Blacklisted thumbnails will not be requested, reducing overhead request count per thumbnail.
Also loading time/requests of 320 result pages will be significantly reduced with viewport only thumbnail loading.
What part(s) of the site page(s) are affected?
Posts/loading
Comments by others
TonyLemur:
Infinite scrolling is a bad idea in most situations for a number of reasons.
- It gets laggy and/or unwieldy when you're many "pages" down because the page has thousands of image elements and the scroll bar is two pixels high.
- There's no way to go directly to a page of results, you just have to keep scrolling and scrolling and scrolling if you need to go back to a certain point in history.
- Related to the above, it's almost impossible to go any appreciable distance back in history, as you're limited to manually scrolling through every single "page" of results instead of being able to click a pagination link to jump back hundreds of pages.
- If the page refreshes, your browser is restarted, or your browser crashes while you're more than a handful of pages back, then you'll have to scroll for a very long time to find your spot again.
Scrydan comment: tumbler does this and I hate it too! Now if this was a user option...eh, don't know. But I like the pagination we have now. Now if it wasn't like tumbler and there was a way to get page number or jump to a point in history with this... this might be slightly better. But there's a reason I stay away from tumbler on my PC.