Nifty, reliable custom software, systems & support. In plain English. Since 1982.
One idea for quick-fix. Not very tested. Use at your own risk. Patch up to latest PHP versions ASAP. …provided Apache has mod_security2 loaded and blocking is enabled. Use: SecRuleEngine On SecRequestBodyAccess On SecRule ARGS_NAMES “@rx (?i)^PHP_SESSION_UPLOAD_PROGRESS$” \ “id:10961801,phase:2,deny,status:403,log,msg:’PHP2Shell upload-progress blocked'” phase:2 lets ModSecurity parse the multipart body, then rejects it before Apache forwards the … Read more
Think twice. Cloudflare’s “prove you are human” wall is not a neutral speed bump; it is an involuntary inspection booth wedged between readers and the web. Their own Turnstile privacy addendum says the system processes “client-side signals” including IP address, TLS fingerprint, User-Agent header, sitekey, and origin, and that the same addendum applies to Cloudflare’s … Read more
A great use of AI and open-source. Deeply, thoroughly open-sourced in fast, solid RUST language, https://github.com/storytold/photocraft is clean-room reimplementation — recreation of the workflows, muscle memory, panels, formats, and habits without copying Adobe’s proprietary code/assets. In other words: not “piracy,” but far more humiliating: A public dashing of its empire as mechanically reproducible. Adobe is … Read more
Just received this email from OpenAI / ChatGPT: “Starting October 30, 2026, your included usage in ChatGPT Work and Codex will decrease from 20 times to 10 times the ChatGPT Plus allowance. Your GPT-6 Pro limit in Chat will decrease from 200 to 100 messages per week. Your price remains unchanged and you’ll keep your … Read more
For white balance, the key is not “gray” specifically. The key is neutral and not clipped. A black, gray, or white target can work if it is spectrally neutral and properly exposed. Gray is popular because it lands in the safest zone. A white card is intuitive for white balance because you’re saying: “this should … Read more
It is said that AI is only: ‘based upon your “training” (which is: discerning token-interconnection likelihood), show what is the next most likely token?’ That is by definition recall. If you “trained” AI solely on the single token “ai is ai” then its answer could only ever be “is” or “ai” or “is ai” or … Read more
SOURCE: https://www.lennysnewsletter.com/p/how-to-turn-your-ai-into-a-world AI models are capable of amazing creativity, but that creativity gets stifled by how they’re trained. Large language models are next-token predictors: at each step, they look at a sequence of text and predict what comes next based on millions of examples. The results may be rated by humans, and those ratings fed … Read more
Precise, fast lookups from massive memory.md It’s clear that a new kind of multi-faceted, multi-layered “memory system” is required to properly, fully serve the LLM harness. Not flat files, not semantica, not graph. What? To my mind, after looking at it for 5 minutes to eval if it’s worth a try …. qmd — ” … Read more
However, AI does love and glean from hints in whatever form. Think of cipher “salt” or the “Heil Hitler” trick / hint / shortcut reportedly discovered in the hacking / breaking of German ENIGMA code machine / cipher. Giving AI a nudge, a starting point, an heads-up remains useful. SimonScrapes covers 20 similar tips. 1 … Read more
