GWGBut it goes back to a question I asked yesterday. Does it matter if you build a better mousetrap if it still catches mice? AKA If I implement the same functionality differently, did I accomplish anything?
petermolnar!tell snarfed I had a brave thought of posting likes/comments to FB via mbasic.facebook.com, disguised as a browser; the problem with that is that it requires username/password, but it might just work
sknebelaaronpk: for some reason my response to your beach-pictures shows up via bridgy, but per WM only as a like not a reply. can you take a look what I did wrong?
aaronpki'm considering switching my code to detect multiple properties, but I think right now most people assume a post is only one thing when receiving webmentions
petermolnarI'd like your (as in #indieweb) opinion: I was wondering if micropub should be possible by posting a single image, and only a single image; no content, no title, no nothing - because all of that is in the exif/iptc/xmp, in the image. I know technically this is not an issue, it's just a question of should this be 'official' or just my own hack
aaronpkit's up to your endpoint to extract data out of the image if you want. for example my endpoint adds location data and sets the timezone from my external GPS logs.
ben_thatmustbemeit all comes down to really what you are doing with your endpoint, is your system creating photos with metadata, or are you creating posts with just a single photo
Loqisnarfed: petermolnar left you a message 5 hours, 35 minutes ago: I had a brave thought of posting likes/comments to FB via mbasic.facebook.com, disguised as a browser; the problem with that is that it requires username/password, but it might just work
aaronpki think one of the earlier iterations enabled server A to get a token at server B without being initiated by B, but it was a lot more complicated
aaronpkallowing unsolicited token generation also has a lot of other security considerations that don't apply when the webmention sender initiates the token generation
gRegorLoveSo I'm just wondering about after the code expires, e.g. what if the receiver doesn't exchange it for a token right away. Guess they need a new code.
gRegorLoveAnd similarly if they do exchange it for a token, they can only read the post for the day or two the token is valid (if we follow the recommendation)
aaronpkmy thought is that by keeping private webmentions relatively simple to implement, we can move forward on a lot of interesting uses for it. whereas trying to spec out full access control and token management it becomes much more daunting and challenging to implement.
Zegnat[benatwork], what UA would you expect? My rel-me verify extension sends its own name as UA, that seemed to be an OK compromise and gets me 301s from Twitter’s t.co links.
[benatwork]Without going into _too_ much detail, on the last two systems I’ve built I’ve wanted humans to go through an auth redirect, and non-humans to not bother with that process
aaronpkif it's about sessions, can't you use the presence of a session cookie to trigger the dance, whereas bots would not be making the request with a session cookie?
[benatwork]No, because in these settings we’re dealing with multiple domains which can’t share cookies with each other, which is usually the reason for the dance
ZegnatKevinMarks, I can’t come up with anything. Some new stuff like private webmentions came out of IWC Brighton, but not much I recall from the demos that can actually be shown off
[benatwork]This is awesome. I’d say the privacy indicator is my biggest tension - ideally in a more-than-binary way, so we know if other people can see the post, what we can do with the third-party representation, and so on. But it’s a huge step forward.
aaronpkYep. the way I see it, the privacy indicator issue is analogous to how webmention at its core doesn't tell you what to do with the post once it's verified