After native XSLT: where should the transformation run?

Started by Anoncto@anoncto••0 views•1 replies
A
Anoncto@anoncto
VIP
18
0
#1

An embedded device can keep returning perfectly valid XML while its browser-facing page stops being useful. If the readable view depended on native XSLT, “replace it with JavaScript” skips the hard bit: who can actually change the device or the client?

The useful migration question is where the transformation should live, and who will maintain it.

XML is not the feature being removed

Chrome's XSLT removal and migration guide targets Chrome 158 (17 November 2026) for removing native XSLTProcessor and XSLT processing instructions, with temporary Origin Trial and enterprise-policy exceptions. That is the published schedule, not a compatibility result for every installed browser.

The same guide explicitly retains XML parsing and XML styled with <?xml-stylesheet type="text/css" ... ?>. CSS may be enough for a simple readable view of existing elements; it is not a replacement for an XSLT stylesheet that constructs a different HTML document.

Choose a boundary you can maintain

Transform before delivery. If you control the producer, render HTML on the server and keep XML available for machine consumers where needed. A gateway can do the transformation for an immutable device, but then obtaining authenticated data, resolving resource URLs and deploying that gateway become your responsibility.

Keep an explicit client-side implementation. Chrome documents JavaScript implementations and a WASM polyfill as migration options. This can preserve a client-side workflow when you can update the application. It also turns the transformation engine into an application dependency: choose and maintain a particular implementation, and compare your actual stylesheets and output. A familiar API name is not a compatibility test.

Manage a compatibility client. For raw XML from devices that cannot be changed, the guide specifically identifies a browser extension as an option. That avoids changing the producer, but every relevant client now needs the chosen extension and its updates. I would review its permissions and deployment scope rather than treat “install an extension” as the end of the migration.

Temporary browser exceptions can buy time. They do not settle which of those components you want to own afterwards.

Preserve the workflow, not just the screenshot

Before choosing, take one representative XML document and its stylesheet and write down:

  • Who can change the producer, the presentation layer and the browser fleet?
  • Which authenticated requests and relative resources are needed to produce the view?
  • Does the result still need to work offline, remain printable or support existing links and scripts?
  • Who owns updates to the replacement engine, extension or gateway?

Then compare the candidate's resulting document and required behaviour, not only whether the first page looks right.

My preference is server-produced HTML when I control the service and ordinary browser viewing is the goal. For an immutable device, I would start with the client-deployment and authentication constraints; a gateway is not automatically the simpler choice.

For a read-only XML endpoint on a device whose firmware you cannot change, would you maintain a narrowly scoped compatibility extension or add a transforming gateway—and which concrete constraint decides it?

0

Join the conversation!