Dennis Snell

[Dennis Snell @dmsnell23]

Can you persuade an LLM to generate tokens until it fails? #llm

https://t.co/SIpUVXUnBG

[Dennis Snell @dmsnell23]

@brendt_gd Talking about HtmlDocument? DOMDocument is full of security holes, corruption, and cannot parse HTML (only XML).

WordPress’ HTML API was built as a performant, streaming, compliant HTML parser easy to embed without WordPress.

Happy to chat if you are interested.

[かんた🌊SWELLマニア @swell_mania]

ボタンブロック内のa要素にdownload属性を付けるなら普通にWP_HTML_Tag_Processor。プラグインは不要です😉

[Dennis Snell @dmsnell23]

the document is the database 🤔
#sgml

[Dennis Snell @dmsnell23]

@ryanwelcher Amazing. Thank you so much @ryanwelcher

[Ryan Welcher | See you at WC Asia in Mumbai @ryanwelcher]

Inspired by Dennis Snell's talk on #WordPress Playground at #WCUS2024, I put together a #VSCode extension to edit ReadMe files in the block editor.

[WPLift @wplift]

🌊 "We're so used to WordPress development being a drop in a massive ocean, but with Playground, *we are* the ocean. We get to choose everything that goes out to our customers," said @dmsnell23, sharing unique opportunities with Playground at #WCUS. 💻✨

#WordPress #PlayGround

[Dennis Snell @dmsnell23]

@laxmariappan @phpcampers Thanks for clarifying. Exciting stuff coming soon in the HTML Processor, doing what PHP 8.4's Dom\HTMLDocument won't even be able to do: fully isolate inner HTML

[Dennis Snell @dmsnell23]

@laxmariappan @phpcampers I don’t understand what the meme is meant to communicate, but I’m curious 😅

[Dennis Snell @dmsnell23]

want to highlight your #wordpress #wordpressplayground based app at #wcus? send me a message!

[Dennis Snell @dmsnell23]

Even spec-compliant #html parsers can "break." HTML cannot represent all possible (even invalid) DOM trees. #wordpress' #htmlapi speaks HTML and ensures full encapsulation and isolation when manipulating a document so this doesn't happen.

[Dennis Snell @dmsnell23]

The #wordpress #htmlapi knows #html so you don't have to.

[Dennis Snell @dmsnell23]

#wordpress #htmlapi knows how to convert #HTML into #XML/#XHTML (but you probably shouldn't do it!)

[Dennis Snell @dmsnell23]

You can't parse #HTML with #XML. Because HTML can't be parsed by XML. XML is not a tool that can be used to correctly parse HTML. The use of XML will not allow you to consume HTML. XML is a tool that is insufficiently sophisticated.

Have you tried using an HTML parser instead?

[Dennis Snell @dmsnell23]

@HowellsMead @nickmdiego It's a repurposed XML parser (cannot parse HTML properly) and unaware of HTML's rules. it's vulnerable to many attacks based on these omissions. it detects tags inside TEXTAREA, it's missing hundreds of character references, unaware of TEMPLATEs, removes content, and much more.

[Dennis Snell @dmsnell23]

SGML lets me write post tags the way I think about them, and with a proper ENTITY and SHORTREF+USEMAP, it parses them structurally.
#sgml

[Dennis Snell @dmsnell23]

@rossmorsali @stevejonesdev @nickmdiego @BlockVisibility oh I could totally see that adding latency. we're exploring adding a final HTML-processing pass in WordPress.

also: the HTML API doesn't _yet_ extend documents but it's designed to stream, meaning output buffering with a callback and no additional latency.

[Milana Cap (#WCEU & #WPDDNS organiser) @DjevaLoperka]

Did you know that HTML API is built from 0, specially for #WordPress, and there's no other framework or CMS that has anything similar? They all still use regex 🥲 There's also an interesting effect on performance 👯

I'll show you its magic in two weeks at @PHPSrbija #PHPSRB 🥰