Dev Notes 17: Blog meta notes and poking at reddit bugs

Dev Notes (DN) documents progress towards other articles. Discussion is preliminary

Today's notes cover brief explorations with a few topics. Mostly this turns into meta commentary about the blog. These do not make for the most interesting findings.

I maybe broke Reddit?

This is a hyperbole. But after much delay, I posted my Part 1 of the IMDb post on Reddit. In relative terms it went well, and while it is not there yet, it might be on track to be the most read post on the blog. However, I also somehow broke the Reddit preview links set via the opengraph tags. These are the little images that show up next to links. I originally set this as a webp for the preview image. This seems to work fine on most platforms, but when I did a test post on Reddit it didn't seem to work. I then switched it to a JPG, and this seemed to work. Well, it works in some cases. In some views it renders fine, but on others it shows up as a broken image for some reason. I had thought it was possible it is trying to load the cached webp image, and serving it as a webp, but this is failing. This theory is supported by a URL parameter in the image it is serving for the preview that has "auto=webp" in the parameters. However, it does not appear to be a malformed webp image. When trying to download these actual images, I get an XML error saying access is denied, so it likely just doesn't exist on their CDN server. It is unfortunate there is no clear way to flush the cache of these images.

I spent a fair amount of what would have been blogging time replying to comments on the blog, or thinking about comments / directions on future posts. It is still an open problem of how to deal with comments and how much time to spend there. I also still need to add features to integrate comments into the bottom of the blog.

I also spent a lot of time thinking about some analytics for the blog. I don't use Google Analytics on the site since it is kind of creepy and a bit disrespectful to the user to participate in Google tracking them everywhere with little actual marginal gain. I use Goatcounter which gives rough numbers in a less creepy way, but might try to recreate some of the analytics on GA in a privacy preserving way.

I probably should just make a newsletter

Yesterday in DN-16 I discussed my reluctance to set up a newsletter. I find them annoying across the internet. Yet, I had some people say they are looking forward to part 2 of the IMDb series, and I don't have a good answer for them on how to find out when it comes out right now.

I spent a little bit of time poking at substack to see just how annoying it would be to mirror my posts. It is still unclear to me how well one can do interactive embedding of iframes, or what the restrictions are. I am leaning towards something like Listmonk, but I might get over that and just use substack to be closer where there already are users.

Other Things

I made limited progress on the Wikidata study. I might try to make some progress on this later tonight as other activities settle, but this won't make the cut for my daily posting goal.

Conclusions

I made limited progress today on new projects. Still there were meta learnings about the blog and some quirks of the web still were useful.

In theory tomorrow I will be focusing on a non-AI/tech project.