Strange. In the last few years, I’ve abandoned search engines for most programming related queries by chugging along happily with Recoll. If I had the time I’d sit down and hack out a web front–end for its Python API but the desktop interface supplied with HTML from my own web crawlers works beautifully (medoc92/recollwebui also exists).

It seems that.. from talking to a few people via email, basic feed discovery with rel alternates are ineffective. Simple redirects work better for organic discovery.

1. /atom -> /path/to/real/{atom,rss}/feed
2. /rss -> /path/to/real/{atom,rss}/feed
3. /feed -> /path/to/real/{atom,rss}/feed

Profiling is the nice, cozy, and lazy way to efficiency. There’s something neat about avoiding optimization and having the numbers to prove it relative to an initial baseline. In PHP there’s Xdebug and KCachegrind and I recently saw php-spx in my feeds. A profiler in any language environment with snapshots of performance over time is just — well nice. The terribad situation is not understanding exactly how something became slow. It might be too late then.

I heard a long time ago.. that poisoning the crawlers is as simple as adding a smiley to the end of every other post. This is satire I promise! :-)

The plant disease identification/management handbook by Balaji Aglave is excellent for popular plants. A lot of modern handbooks are fluffy with information (maybe that’s popular) but this one gets straight to the point — I was very lucky to come across this book a while back.

One aspect of blogging that I don’t like is the unpredictability of an audience’s impressionability. Many people out there read/watch jokes/spam/falsities/uncertainties with utmost confidence — seen it countless times.

Presenting info in a way that prioritizes “critical thinking/reasoning” over an “oracle of truth” is actually hard. I’m almost out of my 20’s and seeing irl reactions to stuff online makes it seem like I’m still in high school — moreso now than ever before. The money gets made somehow :-)

Messing around with a statically generated site can easily lead into a web/browser spec rabbit hole. And… that’s when I remember exactly why everything ends up written outside the browser’s framework into a js framework. I think Firefox is still the only browser that allows easily setting image fallback styling completely with just CSS?

Hyperscript Tagged Markup (htm) is pretty good. It uses tagged templates.

This AI stuff is kinda exciting.. in a “watching danger from afar” kind of way. What kind of feedback loop does it have? Does it finish off the Internet content–wise? The data has to be huge and ultra fuzzy — so someone has to add semantics/structure right or is it automatic?

The Internet is already gamified to an extent… but can it completely auto–generate videos? What kind of exploits will be used against the input? So many questions. It’s like the ultimate frankenstein pandora’s box thought experiment of the most bizarre outcomes :-)

In the end my blog here ended up as a public note taking system — oh well. It’s kinda prickly though because I’m linking out in most posts and don’t have a workflow for auto–archiving the destinations yet.

Chances are that when I return to older posts links inside will be dead. Archive.org isn’t infallible – ya better locally archive anything remotely interesting. Y2Z/monolith is good at creating self–contained web pages.