2021-02-09, 12:03 PM
(This post was last modified: 2021-02-09, 12:17 PM by Laird. Edited 2 times in total.)
(2021-02-09, 11:59 AM)Eldenroot Wrote: I have been testing for 3 days, seems to be fine. I have some ideas but need to write down on paper.
Thanks for testing, my friend! Will be very interested in your ideas!
As I mentioned privately to @Eldenroot (although not framed quite in this way as conflicting), there are two features that I want to add, one of which is for Link Helpers to be able to say "I don't need the contents of the page to know whether I can Help nor to generate a preview" (and thus, by refraining from querying the page's web server, to save on bandwidth), but the other of which is to say "I need the contents of this page to Help". I (we?) need to think carefully as to how to resolve these competing interests.
An example of the former: the existing Tweet helper can generate a (Javascript-heavy) preview based solely on the (last part of the) URL - the ID of the Tweet - without needing to load the tweet page itself. This Tweet helper could then signal "I don't need the contents of the page to know that I can help, and nor do I need them to know how to generate a preview", thus saving us bandwidth.
On the other hand, what about a PDF previewer? This Helper definitely needs the contents of the page (and its Content-Type header in particular) both to know whether it can Help as well as to Help (extract a title and image, as well as an initial preview sentence of the contents from the PDF).
But there are cases where it's not easy to decide which of these two have precedence, and whether we ought to spend the bandwidth and download the page, and if so, on what basis we do so. This is the area which I am mulling around in the back of my mind.