Chris Blume

@programmax.net

Retired software engineer. Volunteering as chair for the W3C PNG Working Group. Formerly Google & Twitch.

Question for coders using AI: Claude seems to be updating AGENTS.md with every change--like a conversation history. I understand updating AGENTS.md when it really should include some knowledge. That's the whole point of the file. But it seems overzealous. Is this normal?

I don't want to like AI vibe coding as much as I do. This grabs all US weather from the NWS and then more actively checks on areas with bad weather. A visitor can of course check their local weather. It does the matching locally against the NWS data it already fetched.

Bild

Thinking out loud: I think AI is a huge problem for big tech. Previously, the difficulty of making all that tech acted as a moat. Big tech now has huge costs and their differentiator is marketing/brand. They'll reduce costs, which will weaken the brand. Thoughts?

I'm curious to make a game but facing existential crisis. I began gaming as a young, shallow kid. Presentation meant more than depth to me. Doom exploded, so I think many others were in the same boat. But today, insane graphics and sound isn't so impossible. It's normal. (1)

I've seen about 9872345 posts about destroying books for AI. I feel the need to chime in because it seems many are missing a key point. They *tried* to preserve the books. The courts didn't let them. The cannot train on it AND preserve it. Be mad at the courts.

The bottleneck was never been code production—its complexity management. This is even more true in the age of AI, btw. AI is good at predicting the next word. But it is not good at understanding meaning. Hence "strawberry". If you can keep complexity low, you can help AI make sense of the code.

I'm curious about the thinking between tying Visual Studio and the documentation together. Old MSDN required you have the matching VS version installed. Was it because you would need a document viewer? And that's how they kept you updated? Was there some business reason?

Once upon a time, I could Google a question and find an answer quickly. Meaning it was in the summary or the page(s) were to-the-point. Showing an ad once was sufficient. They made their money. Now, YouTube needs 15m videos that are mostly filler, sites show multiple ads...

<me> "You aren't bad at math. You're out of practice." Then I read this white paper and think "I'm bad at math." "Each anisotropic Gabor primitive is a Gaussian envelope with a cosine planar wave. A plain Gaussian is a special case of the anisotropic Gabor where the modulation frequency ω is zero.

Holy cow. I forgot I have this. Enterprise Edition, still in shrink wrap. BUT the shrink wrap crushes the box lid over time. I think I need to unbox it. I almost don't want to.

BildBild

I'm trying to get local AI looping. > Ask it to write a complex project. Done in 30s. "That seems wayyyy too fast. No shot." *project doesn't even build* > Ask it to fix its errors. It's been churning for over an hour already...

Earlier AI models were pretty bad at coding. I learned to distrust it and carefully review its work. I need to unlearn that. I need to trust the AI writing it and have a separate AI reviewing it (which I also trust). Otherwise, it is still bottlenecked on the speed of a human.

I've been looking at a bunch of tiny Win32 programs. Almost all of them do not run. It's shuffling bytes around so "according to these 5 people's claim, this is still valid". WinXP even says "This program is not valid." (Compared to I got "This program IS valid, but..")

I thought up a kinda work-around-y way to do this. Commit the docs. Then next commit, remove them. So that "Commit Docs" directory works and no name collision worries. Well, sort of. Same name = change. But that's probably fine. I don't love it. Feels hacky. But I don't hate it

Chris Blume@programmax.net · 2mo ago

Git commits should include files just for themselves. For example, suppose I commit with documentation of the performance testing I did to justify this change. I guess I could just make a "commit docs" folder but that sounds like a mess. I don't want to worry about file names.

Git commits should include files just for themselves. For example, suppose I commit with documentation of the performance testing I did to justify this change. I guess I could just make a "commit docs" folder but that sounds like a mess. I don't want to worry about file names.

I'm writing C++ for a pre-C++11 compiler. It is wild how often I go "maaaan I miss auto". I also ended up writing my own variant & expected. Anyway, it is a clear sign the language is better now.

TIL C/C++ initialization can reference the variable itself and use it for offsets. struct serialized_tree_node{ serialized_tree_node* left; serialized_tree_node* right; }; serialized_tree_node foo[] = { {foo+1, foo+2}, {NULL, NULL}, {NULL, NULL}, };