/blog

Brandon wrote that he likes blogs which do things a little differently and used mine as an example. Of course, I'm flattered that he should talk about it in such terms.

I never deliberately set out to be different per se, there wasn't a day when I suddenly thought "to hell with convention, I'm going to do something else." No, I just started doing what felt right and I made no secret that it was inspired by Dave Winer's blog.

It's nearly six years since I first employed the concept of a 'Today' page. It just felt natural to read the day down the page, just as we do with books. If we're calling it a page then we might as well continue the metaphor.

It just made sense.

The logistics may have changed a bit since but the idea is still the same.

More recently, however, I have consciously wondered about how blogs can be shaken up, and how things can be done differently. But blogs have survived and thrived because of their very nature. There's not that much you can do with the format.

What matters most is that blogs exist, no matter how they look, and that people are continuing to write on the web – sharing their thoughts and ideas with whoever wants to read them.

Using the code examples I got from ChatGPT a little while ago I have now rebuilt the webmention endpoint, at least the part of it that parses the microformats and supplies the required info like name, comment and whether it is a like or a reply.

I was previously using manual lookups by path but now use a recursive function to find values within an array based on the key name provided.

It's a lot simpler, has reduced the amount of code required, and should be more reliable when parsing microformat data from a range of sources. Initial testing looks good but I'll be keeping an eye on things to ensure I've not missed anything.

Yes, the current generation of 'AI' tools can give wrong answers while giving the impression that they are correct, and are limited to the datasets they are trained on but, if you go in with your eyes open, they can be a very useful starting point. Trying to get the same information from a search engine, with pages of links, can often be confusing and overwhelming.

Typical, I forgot to account for the peculiarity of my own site where post links use a URL parameter to distinguish themselves from others on the same day. I had this in the original endpoint but forgot to bring it through to the new one. Fixed!

It's funny how opinions change. A while ago I would have been in complete agreement with Manu with regards post excerpts in an RSS feed:

But the entire point of having an RSS feed is to prevent me from having to actually go to your site to read what you wrote. RSS is not a notification system. It's a distribution system.

That's not technically correct. An item element in an RSS feed has to have either a title or description tag but does not have to include the full text of the item itself.

Since I started using the updates view in /reader the way I read the sites I subscribe to has changed. In this regard I am using RSS as a notification system. Yes, I have the fallback option of switching to the original view and consuming items directly within /reader (should I leave some unread) but this is now no different to using a 'read later' service.

I'm still of the opinion that it makes my reading more interesting, that I invest more in each site, and benefit because of that.