gRegorLoveThe bearer token coming back from the micropub client (like quill) will have one or more scopes requested. Your endpoint code needs to check that.
gRegorLoveOh, and an explicit p-name around just the note's content will be good too. p-name is used a lot in reply contexts. It's implied if the property doesn't exist, so usually returns more than desired. That's why my reply-context to you on https://gregorlove.com/2017/05/welcome-aboard/ has the extra "published" text and webmention stuff
jeremycherfasBut here’s the odd thing. Although I have several routes and paths set in the webmention.yaml, prompted by your question I did a dump on those. And only the first one of each is present in the array. Now that really is very odd.
petermolnarI see Grav popping up more and more, interesting; the last time I tried it it collapsed under my site, I wonder if the media cache handling got any better
jeremycherfasPetermolnar: my site isn’t very media heavy, so I can’t really tell you. I’d love to fix this problem, though, Without it, Grav is useless for indieweb.
petermolnarZegnat for Grav, the used to use an image resizer which was GD based and therefore it wasn't possible to keep metadata for the downsized images, which was a dealbreaker for me
ZegnatI have done Imagick work for the Indexhibit CMS during a summer internship, working on code to keep colour profiles intact or optionally apply the profile and convert the colours in the image so nothing looked “washed out” in browsers.
jeremycherfasPetermolnar I am trying to improve it in one way, which I started at IWC last weekend. But then cweiske identified this other problem which I thought would be reasonably easy to solve.
ZegnatCurrently it does a strpos to find '</head>'. Pull that out into its own variable. Then check that variable and if it didn’t find </head> it should not append the LINK element.
ZegnatAlmost. He uses strpos() to find where in the output </head> is. That is fine. strpos() will return boolean FALSE if it can’t find what it is looking for. It just so happens that FALSE == 0 so the original substr_replace() kept working even if no </head> was found.
ZegnatI do the strpos() earlier and explicitly check if it is FALSE. Because when it is, we are (probably) not dealing with (valid) HTML and we can exit early.
ZegnatIf </head> is found I do exactly the same code he was doing to insert it. I just do a miniscule time saving by not running strpos() again within the replace and use my previously stored variable instead.
jeremycherfasI did try inserting a test for the format of the page, which was why I was asking about return and exit yesterday, but that didn’t seem to work for me.
ZegnatThis just has the big problem that it will still insert the LINK if (for whatever odd reason) you would try to create this JSON: {'headsClose':'</head>'}
ZegnatYes, I might file a PR for this anyway, even if it’s not strictly correct still. You really need to know whether Grav is writing HTML at the time or something else for this to work :/
sknebelif that's really hard to do it might be better to just switch the default to Link-headers, they don't hurt if they end up on the wrong files at least
jeremycherfasZegnat: I tried that, using page.format, or something like it, but I couldn’t get the exit conditions correct. There is a variable that tells you whether the template_Format is HTML, or ATOM or RSS
jeremycherfasAlternatively, and probably better in the long run, download the grav skeleton from getgrav.org and create a user for yourself. Then you can probably download my user folder too.
Loqi[edhelas] All the public Movim feeds can be exported to Atom XML feeds https://nl.movim.eu/?feed/pubsub.movim.eu/Movim . Is his format an IETF one? I prefer to stick to Internet standards and not use some indie protocols and formats for now (especially regardi...
LoqiGrav is a flat-file CMS built on PHP, with Twig templating, and YAML + Markdown for storing articles (YAML for metadata, Markdown for the content) https://indieweb.org/Grav
Zegnatalso, jeremycherfas, maybe add the accounts folder within user to your .gitignore so you don’t check in information like email address and hashes password
jeremycherfasI tghink I see what has happened. I have a separate repo for the plugin alone, because ultimately I want to submit the switch to XRay. And of course the folder is also under /user. But I hadn’t pushed /user in a while
[colinwalker]I've got a plugin which creates a meta box. Then a function is hooked in to save_post to update the contents of a post with values from the meta box. It works fine when the function to do this is in the main plugin file but if I move it to an include it stops working.
[eddie]schmarty++ I too am looking at webmentions, but since I live in the slack world I think I'm gonna filter it into a slack room on my personal slack account