This weekend I worked on some things on Seblog, and since I did some productive work on other projects today, I feel like I can write a blogpost about it without sounding so procrastinating.

Inspired by Aaron Parecki's monthly overviews, and triggered by @zegnat, who was looking for the same type of page on my site, I started to 'deconstruct' my URLs. You can now remove parts of it and still get useful pages. That was on my list of itches since november 2015 now.

So let me explain my URL design first. This post, for example, is /2017/06/20/1/daily-and-monthly-overviews. This is an idea I got from the wiki, and probably also originates with Aaron. The identifying part of it is the date, followed by an ID (the nth post of that day). The slug is just for humans and can be omitted: you will get redirected to the right one. Some posts don't even have one. This allows for a Whistle-style URL shortener which I run at 5eb.nl. (This post can be found through 5eb.nl/4ox1.)

The first step was to remove the ID, to get to a day view. I just list all the posts that are created on that day, including private posts when I'm logged in, but oldest-first instead of the newest-first order of my feeds. I also display a summary of the day in icons at the top. I'm really pleased with how it turned out!

After that, I also wanted the monthly view, to get a better overview. I'm really pleased with that too, it's nice to go through my old posts this way, seeing old memories. I'm now more interested in importing my Facebook posts too. I am still looking for a way to have some anchor or summary of the day in my monthly view, to find posts quicker, and adding a location like Aaron might do the trick, but I'll have to figure out how to do that still :)

I'm quite happy with this for now!

The past week I mentioned both Martijn and the Twitter-account of the Dutch Railways (@ns_online) in different blogposts. For Martijn, I used a hand-written link with the proper .u-category.h-card classes to person-tag him. [see update below] For @NS_online, I wanted to @-mention them in the POSSE'd tweet. Martijn complained that my blog didn't autolink them, so that's what I fixed now.

I have a new syntax to @-mention (and thus tag) people in my blogposts.

• I want to match names like @name
• I don't want to match the word @-mention itself.
• I want to be able to escape the @-mention with a \, like \@name, so I can talk about @-mentions in a blogposts (this one, actually, I like meta-meta-meta).

Then the syntax. Obviously I use an @-symbol, with a name behind it. I then check the name for the following:

• Is the name on my list of names? Then use the URL I provided.
• Does the name contain a dot? Then assume it's a domain / URL itself, so add replace @ with http:// (@seblog.nl becomes <a href="https://seblog.nl/">Sebastiaan Andeweg</a>)
• And if it doesn't contain a dot, assume it's a Twitter account.

I then fetch the h-card (or profile information) from the resulting URL, and use the name that's there.

There are times, however, where I want to specify the name. In the example at the beginning of this post, I called Martijn 'Martijn', not 'Martijn van der Ven', which is on is h-card. Sometimes a full name makes no sense. So I can add the name I want to use in brackets behind the tag (@Zegnat[Martijn]). This way I have total control over my text (and possible conjugations if I start writing in a language that needs those, not that I speak one).

Then it all comes down to the following syntax:

@namefromcache        > the URL I specified
@someone.com          > that URL
@facebook.com/someone > also the literal URL
@twitteruser          > their Twitter profile URL
@someone[This Person] > my own name for them
\@someone             > escaped tag

I now need to redo my automated webmention-sending, for it does not yet recognise these tags. Manual sending works fine!

Edit: after some discussion during the Virtual HWC EU-time, I realised that just doing @someone is not really person-tagging them. Person tagging is explicitly saying "I am now with this person", which makes sense for photos and checkins, but not always in blogposts.

I added a minor tweak: I can now use + instead of @ to really person-tag someone. The @ is used for just a mention, the +-mention will receive the class="u-category h-card" markup. I don't know how much I will be using that, but I have that option now.

Hi! I saw your like and your re-webmention (because the picture is broken). I think I should've found the right author using all the steps on /authorship – XRay does find it and I'm thinking about switching to using that for my mentions.

What I did see on your side, however, is that you have a p-author outside of any h-*.

I think my site would've found your h-card if it was the first thing on your homepage, or if you would've extended your h-entry to include that p-author h-card. But that only makes your site easier to parse, I should've found it with this already :)

A little while ago I added reacji to my site. Since reacji are just comments who's content contains only one emoji, I did support reacji before the change. I only changed the way I displayed them, when I send them.

One thing I noticed after my return to Facebook, is that the like-button now has several nuances. These love, haha, wow, sad, angry-things are sent by Brid.gy as emoji. Because of this, it didn't took long for me to receive my first reacji via webmention. They just showed up as comments with one emoji.

So today I fixed that:

Note that I already used a heart icon for likes. I want to keep it that way, the emoji looks different enough to distinguish them. If you comment on Facebook with a heart, they display it as a comment with a heart, and I will just show it as a like, like above. But I'm fine with that. You can also reply with other emoji if you like.

Oh, and I fixed the images now too.

Yesterday I re-activated my Facebook, for the first time since 2.5 years. Still not really sure about this silo-return, but I want to try it again. Facebook is a powerful tool, and with the other tools IndieWeb gave me, I think I can manage my relation to it better than I could 2.5 years ago.

At first sight, it might look as if IndieWeb is about stopping silo's. In that light, a return looks strange. But there is no IndieWeb principle that says you should quit Facebook or Twitter. As long as I POSSE, I own my data, and I can still connect to my silofriends.

I still want to minimise the actual time spent I on their site. So today I made the first few bridges that should make that happen: I hooked up by site to brid.gy/publish and the 'normal' Brid.gy backfeed for Facebook. Now I can post via Micropub to my site and automatically post to Facebook too. Any comments and likes received on the post on Facebook will return to my site as well.

Now I have to make some interesting content to test all this with.

Also just realised I have no-one to talk to about this. It feels so lonely, being connected again, because nobody shares that 'again' aspect of it. Everyone I know is on Facebook, or I think they are, and I don't want them to happily poke my newly activated account. I want to see what Facebook does with my account. Will it try to bring old posts under the attention of friends, to give me likes and comments, so I stay? Or will it be unnoticed?

Now I have to go to a real-life event. I will try and resist telling people about my Facebook adventure. (That's why live blogging about it in English seems a good idea. I think this will be unnoticed as well.) The nice thing is: I can just post an RSVP to a Facebook event via my site. Will hook up Brid.gy tomorrow, if I still think Facebook is a good idea then.

You guys have no idea how scary it is to log in to Facebook after 2 years. (I have been back for 15 minutes at the 0.5 year point to transfer ownership of some Pages.) My profile picture is still the one from back then (although the black and white one I use on this site is actually older). There are 97 notifications in the world-icon and I don't want to click them. I scrolled through the feed, a few posts, and they all look like personal insight in lives of people I haven't seen in years. With every move of my mouse I feel like the addiction is right around the corner. But it could be me, freaking out, because I know the countless hours I spent scrolling that page. I have no idea what I want with this.

I'm logged out again. No disabling, just logged out. We'll see tomorrow.

Yesterday, I went to Den Helder for a day. Besides doing some work, we also made a walk through the dunes. Such walks produce pictures, somehow, and I wanted to post some of them here.

Since I didn't want to create multiple posts, it made sense to create a list of photo's in one post. Instagram also launched a new post type with multiple photo's, and this seemed to be a good match for it.

So yesterday I made a quick foreach loop to be able to post them and today I styled it some more with CSS. It's not as pretty as the Instagram presentation of a multi-photo post, but I didn't want a carrousel like they use. The pro is that the images are viewed one by one, not influencing each other that much, but the con is that it's hard to signal to the user that there are more then one picture behind the post. I like the list as it is now. It's kind of long, but it works.

Today, I moved some existing code from my kirby-micropub plugin to my indieweb-toolkit, and then I rewrote it a bit. It is the code that checks for an Authorization: Bearer xxx header and rejects people that have the wrong token.

At this point, my blog just uses tokens.indieauth.com, as it's token endpoint. So, that is what the toolkit uses now too. This is not ideal, and I plan on adding a token endpoint directly to the toolkit. But, everything in steps!

You can now do the following:

indieauth::requireMe();
indieauth::requireScope('create');

// do stuff!

And then the script will exit if there's no Authorization: Bearer in the header with the proper scope and a 'me' value corresponding with the current URL. (You can also pass in a 'me' to check against.) I am still not sure about some things, so I am putting them out here to think about them some more. Feel free to comment.

• Is 'IndieAuth' the right name for this static class? I think so, because it uses 'me' and 'scope'. But at this point, it's only checking tokens. And when I add a token endpoint, is that token endpoint still IndieAuth?

• In my code, I now check url::host($token->me) == url::host($requiredMe)). I only compare hosts, so seblog.nl, which works, because seblog.nl/micropub still has the host seblog.nl. Maybe I should drop this 'use the current URL if the $requiredMe is empty' and only go for explicit 'me'-values.

• Previously, I threw Errors. Now, I just set the HTTP-header and exit the script. I wonder which way is more elegant. The way I do it now, makes sure the right HTTP status is sent, but the way I did it before allows for more customisation. Both ways exit the script, which is the most important part.

I changed my mind about reposts, again. This is what we call voortschrijdend inzicht in Dutch.

A few days ago, I changed how I display reposts. In an attempt to honor copyright, I only display the first 50 words of a repost. This way, it's more a quote than a full copy of the content, but I can still retweet stuff, since cramming 50 words into 140 characters is doable, but rare.

What's more: I only showed the repost in the feed, not on my permalinked pages. I redirected those permalinks to the original permalink. This way, while I had the original content imported on my site, I wasn't actually re-publishing much of it.

But today I reposted a much reposted post of Aaron, and then it stuck me: I can't send a webmention, in the 'oh look a comment' kind of way, if I don't have a permalinked page with Microformats on it.

In a way, a 302 redirect has the URL of another page on it by using a HTTP Location header, so it does link back to the original page. The question is then: is that a valid link for sending a webmention? And: does a 302 redirect on the source URL to the target URL imply a repost of the target? Not really, I think. It's just a redirect.

So I got rid of the redirect of my permalinks, giving my reposts their page back. I still only display the first 50 words of a repost, making it more a quote than a repost. But aren't reposts just full quotes? It's still the major part of the content of the post. I will keep it that way for now.

Edit 20:17 CET After some discussion on IRC, I decided to also ditch the 50-word-cutoff. Only showing the first 50 words of a repost makes it a quotation, not a repost.

The need for the 50-word-cutoff was born because I felt uneasy reposting a whole article. The solution to that is: don't repost a whole article. The problem was that I already reposted a whole article, because at that time, I found it important to keep it online, in case the original server would collapse under a lot of trafic. The next day, that wasn't a reason to keep it on my site anymore.

So, I changed that single post to a bookmark (replacing only repost-of to bookmark-of in the .txt of my post) and now I'm good again. I removed the code that showed only the first 50 words. If I ever want to quote something, I can use quotation-of, but for now, I'm fine.