YHK - yet another hacker news clone

Chrome and Edge are removing XSLT: what breaks and what does not

51 points croes 2 hours ago 57 comments

tossit444 2 hours ago | parent

503

Insimwytim 1 hour ago | parent

So, this site breaks, we know that much

Barbing 1 hour ago | parent

doctor-seo dot net has failed us. What is this world coming to?

troupo 2 hours ago | parent

There's a long discussion about XSLT removal here: https://github.com/whatwg/html/issues/11578#issuecomment-595... (the link is to the comment with this proposal https://github.com/doriantaylor/swet which already is much better than any templating proposal "the platform" is trying to push)

pjmlp 2 hours ago | parent

Edge is Chrome.

zamadatix 47 minutes ago | parent

The article says as much in the opening. It also includes a statement from Microsoft noting they do not intend to carry this forward separately, with a salient comment in one of the Microsoft page citations:

> For functionality and compatibility reasons, Microsoft Edge adopts nearly all of the Chromium project's changes to the web platform. However, Microsoft retains full control of the Microsoft Edge browser and may defer or reject changes.

Removed features don't always follow the same timelines either (and not just because of a lag in syncing to upstream Chromium).

Trustable8 1 hour ago | parent

Shame back in the XHTML days XSLT was a spec that could have made website modularity so much cleaner. I once created an early website creation tool that merged separate components using XSLT it was really powerful in the days before javascript was usable without jQuery. However XHTML was dropped in favor of HTML5 and XSLT was pretty much useless after that. Its done well to survive this long I presume it was mainly kept to style XML files.

g-b-r 1 hour ago | parent

The XSLT 1.0 spec specifically was written pretty badly, though, to my recollection; I had a hard time at making sense of it

WCSTombs 1 hour ago | parent

Yeah, XSLT 1.0 has some really obvious missteps, so much so that to my knowledge basically nobody just implements XSLT 1.0. What they actually support is (most of) XSLT 1.0 plus a handful of EXSLT extensions (which are completely unofficial and not part of any XSLT spec, by the way). In my opinion the EXSLT stuff is the bare minimum needed to make XSLT 1.0 at least a passable language. Meanwhile, the XSLT spec evolved into 2.0 and 3.0, which as far as I'm aware, even to this day, has no complete open-source implementation.

wahern 1 hour ago | parent

HTML5 could have taken the extra step of defining a canonical XML or XML-compatible DOM encoding, or at least encouraged it. But the anti-XML sentiment was explicit and spiteful. Still, popular XSLT processors are built to consume and even produce HTML5, I'm just not sure how thorough those capabilities are or whether they're based on any formal or informal standards. (Maybe the W3C has/had something like that, as part of it's attempts to rollup HTML5 snapshots into versioned standards?)

I lost any enthusiasm for XML after my experience with WDDX and XML-RPC, but as a foundation for HTML it was the best thing going. And maybe still is, given the absence of cross-toolchain contenders? Ditto for XSLT, at least if disregarding the choice to make the syntax homoiconic. In fact I circled back to XSLT long after it was clear it was a dead end. A good tool is a good tool, and the fact it remains so despite the wanton neglect only proves it.

jxramos 1 hour ago | parent

I'm curious, what exactly in the DOM is not XML-compatible?

degamad 23 minutes ago | parent

From long-lost memory, I think it's implicitly closed tags (e.g. img).

jdiff 5 minutes ago | parent

XML can handle implicitly closed tags, they call them "empty" in the spec here:

https://www.w3.org/TR/xml/#sec-starttags

masklinn 1 hour ago | parent

> HTML5 could have taken the extra step of defining a canonical XML or XML-compatible DOM encoding, or at least encouraged it.

Yeah and then we could have called it xhtml5. Too bad that does not exist. Oh wait https://html.spec.whatwg.org/multipage/xhtml.html

> But the anti-XML sentiment was explicit and spiteful.

No, xhtml was just a bad idea creating tons of problems and solving none.

kllrnohj 1 hour ago | parent

> xhtml was just a bad idea creating tons of problems and solving none.

Well an XHTML parser (and serializer) is a lot simpler, which is a pretty meaningful problem to solve.

What problems does xhtml create?

smallerize 1 hour ago | parent

The main one is that even a slightly broken XHTML document will fail to render. HTML is much more forgiving.

JoshTriplett 1 hour ago | parent

With the benefit of hindsight, if the web had been that way from the beginning, I think we would have been better off. The moment you tested your content in any browser, you'd immediately see if it was broken.

But it wasn't something that could get introduced later, after all the broken content that already existed.

Conan_Kudo 45 minutes ago | parent

Yeah, I think SGML HTML as the first pass was a mistake. All my early websites were XHTML until people called them "bad"...

ktpsns 42 minutes ago | parent

With all respect - the opposite is true. The web succeeded because HTML and HTTP are easy standards easily writable by humans with text editors. HTML was writable by non-programmers.

JoshTriplett 27 minutes ago | parent

I don't think "parsed strictly" makes something hard; on the contrary, I think it's helpful to make something easy. Clear feedback, easy to test.

But then, I've always preferred strongly typed over stringly typed.

joshuamorton 19 minutes ago | parent

This imagines HTML as code and not markup.

The reason HTML and markdown are more widely adopted than latex is because the former are more forgiving.

JoshTriplett 12 minutes ago | parent

I am thinking of it as markup. I stand by the point: I think "forgiving" is not a feature in parsing; it's a bug.

leoc 27 minutes ago | parent

That's not obviously so. The adoption of JS wasn't particularly hindered by the fact that users have to match their brackets and generally avoid syntax errors.

bazoom42 41 minutes ago | parent

> Well an XHTML parser (and serializer) is a lot simpler

Perhaps, but a combined html and xhtml parser is more complex than just a html parser. Since html already is entrenched, xhtml would be in addition to html, not instead of.

johnny22 1 hour ago | parent

i think xhtml happened before the wider world of web devs were ready for it. Component based architectures took longer to be adopted than I certainly expected.

It might actually work if reintroduced these days as long as we can consider many components to be valid documents.

bazoom42 57 minutes ago | parent

There is nothing component-based about xhtml compared to html. It is just html with xml syntax.

lofaszvanitt 1 hour ago | parent

And the rendering was very slow back at the day.

Devasta 1 hour ago | parent

It is wild to look back at XHTML and see that the W3C were entirely correct about XML. They envisaged a world of schemas, client side templating, custom elements, dynamic forms, the works.

Alas, for laughably bad reasons we had to scrap all that for HTML5, Javascript and CSS; a stack so awful that most devs doing anything on the frontend immediately scramble to React, TypeScript, Tailwind, whatever it takes to avoid the underlying garbage.

A horrid state of affairs that will only get worse as time goes on.

r_lee 1 hour ago | parent

yet people will say that HTML5 is all you need or something

ebiester 1 hour ago | parent

They were wrong about one thing: tooling matters.

The XHTML/XML/XSLT tooling was bad, and XSLT was hard to reason about. That meant that amateurs had no way to learn it, and those amateurs went to school and became the workforce.

The web became a winner because it was easy to get something working. And then Google made it a mission to make great developer tools in everyone's hands for free.

I think in 2026 we could build the tooling around a better experience, but it would require thinking about the developer experience first.

Telemakhos 1 hour ago | parent

You could do some awesome things with XML/XSL. My favorite was RSS feeds or OPML collections of RSS feeds (both flavors of XML) calling an XSL transformation when viewed in a web browser instead of mined by computers for data; same with TEI files, although that was a bit more arcane. The idea was that you'd have one data file for machine use and a way to present that data to humans as a web page.

But, as a stack, the XML/XSL/XPath/etc combination was weak because XSLT was hard to comprehend and never updated. No browser supports XSLT 2 or 3; they froze support with 1. XSLT 2 was 2007, a year before HTML 5 was a thing, yet it never got picked up by browsers. Javascript was just easier to understand than XSLT. Development of XHTML was also glacial, and people wanted new tags, especially for video.

mpyne 50 minutes ago | parent

> XSLT 2 was 2007, a year before HTML 5 was a thing, yet it never got picked up by browsers. Javascript was just easier to understand than XSLT.

Turns out that tasks easily done by programming languages are best done by programming languages. Who knew?

JSON allowing frontends and backends to communicate in a way analogous to the "smart endpoints, dumb pipes" model for networking saved the Web. The XML adherents wanted the inverted model, which was always insanely difficult to reason about and led to entirely XML-specific security threats.

WCSTombs 1 hour ago | parent

I can't help but agree, and sometimes I think about the alternate version of the web where XML realized its full potential as the dominant technology. The obvious problem with XML is that the syntax is so verbose and redundant that it's only barely human-writable, but this is 100% a solvable problem.

cousin_it 1 hour ago | parent

When I first learned that XSLT was going away, I got curious how to achieve the same thing in JS (client-side templating that happens during page load, without any "flash of content"). It turns out there's a very nice way to do it, by using inline scripts that append elements next to document.currentScript. Lean Rada wrote about it here: https://leanrada.com/notes/inline-rendering-currentscript/ and I wrote about it as well: https://vladimirslepnev.me/write

WCSTombs 1 hour ago | parent

That Lean Rada article is really fun!

krylonmae 1 hour ago | parent

Looks like the site broke

jiggawatts 1 hour ago | parent

The breakage that sticks out to me is that IIS in Windows Server outputs some reports as XML+XSLT and these have become nigh unusable for years now.

I ignore every official statement about “support” and simply look at the horse's mouth: Microsoft hasn’t bothered to fix this in IIS, and likely never will.

It’s a dead product, “supported” only in a contractual sense, not a practical one.

(If you called support right now demanding fix for this, they’d tell you that it’s broken by design, have you tried moving to a recurring revenue service offering?)

yokoprime 1 hour ago | parent

i just got flashbacks to a past life where i handcrafted XSLT to pre-render output from a service as HTML server side. Of course the front end was running some asp.net version and Microsoft flavored XSLT had lots of quirks.

compiler-devel 1 hour ago | parent

In the age of LLMs I would expect that legacy code bet carried indefinitely. The LLMs will have it indefinitely in their training data, and with code "cheap" to produce now, what are the good reasons for dropping legacy code?

anonymous908213 1 hour ago | parent

The obvious answer is that LLMs are not quite the machine gods they're marketed as, innit.

compiler-devel 51 minutes ago | parent

I don't understand what you're saying.

anonymous908213 38 minutes ago | parent

Your presumption is that code is "cheap" because of LLMs. I am challenging the premise. Maintaining code in production for a product used by billions of people is not cheap, with or without LLMs. Perhaps your shitty disposable code that you write for yourself and never subject anyone else to is cheap now, but if your code is in a large, complex codebase and has to actually work for everyone who uses it, not so much.

compiler-devel 32 minutes ago | parent

LLMs are only getting better and more competent, especially for code well-covered by tests. On such codebases it's definitely cheaper to maintain code with a LLM than by humans alone. I presume then that the XSLT implementation in Chrome is not well-tested.

bazoom42 51 minutes ago | parent

> what are the good reasons for dropping legacy code?

Reducing attack surface.

compiler-devel 50 minutes ago | parent

That is a good reason. Is XSLT known for security problems the same way that CGI was?

Velocifyer 52 minutes ago | parent

What will we do for RSS/Atom feeds?

bawolff 49 minutes ago | parent

Not style them?

Use css to style them instead?

dfabulich 36 minutes ago | parent

Style them with CSS. https://danfabulich.medium.com/beautify-your-rss-atom-feeds-...

mcculley 47 minutes ago | parent

I am still using XSLT. I have been using it for 26 years. I generate XML documents on the backend and turn them into HTML for various reports. XSLT has a lot of features and power that newer templating languages lack.

Are any of these newer templating languages worth me switching?

WJW 43 minutes ago | parent

Yes, because XSLT won't work in Chrome and Edge anymore.

Klonoar 36 minutes ago | parent

Their comment isn’t clear if they’re doing the transform on the server before piping it down as HTML, which some implementations have done over the years as it provided access to newer XSLT features that browsers stagnated on.

miohtama 8 minutes ago | parent

You can still run JS or WebAssembly based XSLT transformer

sippingabonedry 16 minutes ago | parent

A great way to convince you you've always been wrong:

Step 1: Gain browser monopoly

Step 2: Generate fake security scare

Step 3: Gaslight everyone that only way to mitigate is to get rid of it completely like Flash.

Step 3a (corollary): There isn't anyone available that can fix the bugs because we spent all resources on UI/UX experimentation

post-it 2 minutes ago | parent

Thank you for reminding me that we don't have Flash anymore. I'm counting blessings I forgot I had.

hmokiguess 42 minutes ago | parent

XSLT is glorious, too bad.

goodridance 8 minutes ago | parent

Good riddance. XSLT and EJB marked the dark ages of software development. For a half decade nothing made sense and programmers and management were going along with the charade until everyone collectively realized it was complete garbage and just stopped using it. I remember replacing an XSLT/EJB system that processed one order per minute with a C++ system written in a week that processed 10,000/minute. A team of a dozen programmers had been working on that XSLT disaster for years.