Support for adding RSS/Atom addresses manually

[jan-kleks]

I have tried to add some pages that I follow regularly, but unfortunately the RSS detection feature often does not work... Manually adding addresses would have been better.

For instance, this page does not work https://twostopbits.com/ or this one https://romhacking.ru/ (I am heavily into retro video gaming news)

[snarfed]

Good feature request!

[FurblandChannel]

A counter-proposal: generating handles for sites that have (for instance) multiple RSS feeds might be confusing, so I think the service should simply check for RSS feeds linked to by / (like both of those sites have). For instance, https://www.pocketgamer.com has a completely fine RSS feed at /index.rss which is linked to on the main page, but not defined via a meta tag. If I were you, I would simply HEAD all the links on the page and check if any are XML and/or RSS and then investigate to see if any are valid feeds from there.

[snarfed]

It's a good idea!...but tricky to get right. There are feed directories out there, so I'd probably need to limit it to feeds on the same domain. And many sites publish multiple feeds, eg all posts vs comments vs specific sections, etc, and I can't easily know which is which.

Less important, but still significant, web site front pages often have hundreds to thousands of links. Fetching them all just to discover feeds would be a lot.

This is what metadata discovery is for. If you're looking for a specific thing semantically, eg "the RSS/Atom feed with this web site's posts," HTML scraping is brittle at best, and often dangerous.

[FurblandChannel]

That's why I’m suggesting to HEAD all the links to discover the Content-Type instead of GETing all of themOn May 3, 2025, at 6:03 AM, Ryan Barrett ***@***.***> wrote: snarfed left a comment (snarfed/bridgy-fed#1884) It's a good idea!...but tricky to get right. There are feed directories out there, so I'd probably need to limit it to feeds on the same domain. And many sites publish multiple feeds, eg all posts vs comments vs specific sections, etc, and I can't easily know which is which. Less important, but still significant, web site front pages often have hundreds to thousands of links. Fetching them all just to discover feeds would be a lot. This is what metadata discovery is for. If you're looking for a specific thing semantically, eg "the RSS/Atom feed with this web site's posts," HTML scraping is brittle at best, and often dangerous. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***>

[snarfed]

Right, I got that. It's nice that HEAD bandwidth is less than GET, but that's not a meaningful cost or burden for us, and otherwise it adds effectively the same CPU, latency, processing, codebase complexity, and other overhead.

It might be a good idea regardless though! At least, if we find a reasonable heuristic for choosing between multiple feeds. I can definitely consider it.