Keith Cirkel
@keithamus.social
Software Cyber Shepherd.
Web platform engineer. Participant: WHATWG, CSSWG, ARIAWG, OpenUICG, WebComponents CG
Website: https://keithcirkel.co.uk/
GitHub: https://github.com/keithamus
Mastodon: https://indieweb.social/@keithamus
Web platform engineer. Participant: WHATWG, CSSWG, ARIAWG, OpenUICG, WebComponents CG
Website: https://keithcirkel.co.uk/
GitHub: https://github.com/keithamus
Mastodon: https://indieweb.social/@keithamus
For example, some elements of XSLT have been broken in several browsers for a good while and only since this deprecation have people noticed.
November 8, 2025 at 10:18 AM
For example, some elements of XSLT have been broken in several browsers for a good while and only since this deprecation have people noticed.
I've spoken to devs who've pushed back on pdfjs, mostly because of the way pdfs are not "first class" so the interaction cab be confusing for their customers, but that's incidental to the extension I guess.
I think I agree though, but the question becomes how often is it used?
I think I agree though, but the question becomes how often is it used?
November 8, 2025 at 10:18 AM
I've spoken to devs who've pushed back on pdfjs, mostly because of the way pdfs are not "first class" so the interaction cab be confusing for their customers, but that's incidental to the extension I guess.
I think I agree though, but the question becomes how often is it used?
I think I agree though, but the question becomes how often is it used?
AIUI Chrome will - upon the user first visiting an XSLT adorned document - suggest the user download the extension. I think there's an opportunity to have the extension built-in but historically users have pushed back heavily on built-in extensions, though.
November 8, 2025 at 10:05 AM
AIUI Chrome will - upon the user first visiting an XSLT adorned document - suggest the user download the extension. I think there's an opportunity to have the extension built-in but historically users have pushed back heavily on built-in extensions, though.
It seems straightforward for devs to ship a JS version of XSLT, they can pick their own version, and aren't stuck to whatever the browser ships. To me this seems like a win for developers, but I'd like to hear why it's not.
November 8, 2025 at 9:57 AM
It seems straightforward for devs to ship a JS version of XSLT, they can pick their own version, and aren't stuck to whatever the browser ships. To me this seems like a win for developers, but I'd like to hear why it's not.
...but to jump to supporting 3.0 seems like high-effort low-reward considering the total userbase of XSLT.
I'd like to think of these as two separate issues though. Right now we're saying "XSLT 1.0 isn't good for anyone" and the next step is "BYO XSLT" which seems palatable to some, but not others
I'd like to think of these as two separate issues though. Right now we're saying "XSLT 1.0 isn't good for anyone" and the next step is "BYO XSLT" which seems palatable to some, but not others
November 8, 2025 at 9:54 AM
...but to jump to supporting 3.0 seems like high-effort low-reward considering the total userbase of XSLT.
I'd like to think of these as two separate issues though. Right now we're saying "XSLT 1.0 isn't good for anyone" and the next step is "BYO XSLT" which seems palatable to some, but not others
I'd like to think of these as two separate issues though. Right now we're saying "XSLT 1.0 isn't good for anyone" and the next step is "BYO XSLT" which seems palatable to some, but not others
I'm very happy to make the case, and have done so. Browser engineers involved in this have poured over all the comments to do with XSLT and there has been a lot of consideration about what is the right path forward. Continuing to support XSLT 1.0 doesn't seem right; many involved want 3.0 support...
November 8, 2025 at 9:52 AM
I'm very happy to make the case, and have done so. Browser engineers involved in this have poured over all the comments to do with XSLT and there has been a lot of consideration about what is the right path forward. Continuing to support XSLT 1.0 doesn't seem right; many involved want 3.0 support...
Text fragment links are shipped in chrome and Firefox iirc. Do they solve this issue? developer.mozilla.org/en-US/docs/W...
Text fragments - URIs | MDN
Text fragments link directly to specific text in a web page, without requiring the page author to add an ID. They use a special syntax in the URL fragment. This feature lets you create deep links to c...
developer.mozilla.org
November 8, 2025 at 8:26 AM
Text fragment links are shipped in chrome and Firefox iirc. Do they solve this issue? developer.mozilla.org/en-US/docs/W...
There's been movement on this. It's hard to find the right balance of what the browser should do and how it makes it easier for developers, though.
Rest assured we'll keep working on it!
Rest assured we'll keep working on it!
November 7, 2025 at 11:05 AM
There's been movement on this. It's hard to find the right balance of what the browser should do and how it makes it easier for developers, though.
Rest assured we'll keep working on it!
Rest assured we'll keep working on it!
AIUI if you don't have an `&` then it's counted as a child combinator (the ` ` character). So I guess one way to think about it is that `&` smooshes the selectors together, whereas no `&` leaves whitespace - therefore child combinator.
November 7, 2025 at 10:57 AM
AIUI if you don't have an `&` then it's counted as a child combinator (the ` ` character). So I guess one way to think about it is that `&` smooshes the selectors together, whereas no `&` leaves whitespace - therefore child combinator.
To many, it seems. I believe it's being discussed next week! 🍆
November 7, 2025 at 10:55 AM
To many, it seems. I believe it's being discussed next week! 🍆
It's also extremely unknowable for the browser (because of a whole bunch of complexity), and is computed as the user makes the interaction to move to the next focusable element.
There's progress in this direction though, namely: open-ui.org/components/b...
There's progress in this direction though, namely: open-ui.org/components/b...
beforefocus/focusNext (Explainer) | Open UI
Open UI
open-ui.org
November 7, 2025 at 10:54 AM
It's also extremely unknowable for the browser (because of a whole bunch of complexity), and is computed as the user makes the interaction to move to the next focusable element.
There's progress in this direction though, namely: open-ui.org/components/b...
There's progress in this direction though, namely: open-ui.org/components/b...
This is being worked on! The latest iteration is w3c.github.io/gpc/. The big problem here is that legislature needs to get involved, and if you think web standards are slow, wait til you see how long it takes to get unilateral agreement from global governments 😂😭
Global Privacy Control (GPC)
w3c.github.io
November 7, 2025 at 10:52 AM
This is being worked on! The latest iteration is w3c.github.io/gpc/. The big problem here is that legislature needs to get involved, and if you think web standards are slow, wait til you see how long it takes to get unilateral agreement from global governments 😂😭
Right, the polyfill might be viable but it seems of the very few people who are keen on XSLT, there's a portion who don't seem to like the idea of a polyfill.
Even with the polyfill, there are inherent security issues, AIUI, because script can run in surprising places, but I am not fully read up.
Even with the polyfill, there are inherent security issues, AIUI, because script can run in surprising places, but I am not fully read up.
November 7, 2025 at 10:51 AM
Right, the polyfill might be viable but it seems of the very few people who are keen on XSLT, there's a portion who don't seem to like the idea of a polyfill.
Even with the polyfill, there are inherent security issues, AIUI, because script can run in surprising places, but I am not fully read up.
Even with the polyfill, there are inherent security issues, AIUI, because script can run in surprising places, but I am not fully read up.
QR codes are likely something tenable, but I'm curious - they're very customisable, e.g. with center images, colours, rounded corners, the number of check bits, etc. What does an API look for this that doesn't become attribute soup?
November 7, 2025 at 10:49 AM
QR codes are likely something tenable, but I'm curious - they're very customisable, e.g. with center images, colours, rounded corners, the number of check bits, etc. What does an API look for this that doesn't become attribute soup?
On self-links for headings, can you help me understand the problem? I have some guesses as to what this means but I'm not sure.
November 7, 2025 at 10:47 AM
On self-links for headings, can you help me understand the problem? I have some guesses as to what this means but I'm not sure.
w3c.github.io/gpc/ may be a solution to cookie dialogs. TL;DR, it's like the old DNT header but this time with actual legislative weight.
Global Privacy Control (GPC)
w3c.github.io
November 7, 2025 at 10:47 AM
w3c.github.io/gpc/ may be a solution to cookie dialogs. TL;DR, it's like the old DNT header but this time with actual legislative weight.
I think you'd need an `&` there, so e.g.
input:hover {
&::file-selector-button {}
}
input:hover {
&::file-selector-button {}
}
November 7, 2025 at 10:44 AM
I think you'd need an `&` there, so e.g.
input:hover {
&::file-selector-button {}
}
input:hover {
&::file-selector-button {}
}
Developers voices always matter! That’s exactly why we’re here.
Having said that, xslt is… complicated. No one maintains any of the existing implementations and building a new one would be a huge lift for what it seen as a very small user base. It’s hard to justify its place in the web.
Having said that, xslt is… complicated. No one maintains any of the existing implementations and building a new one would be a huge lift for what it seen as a very small user base. It’s hard to justify its place in the web.
November 6, 2025 at 9:03 PM
Developers voices always matter! That’s exactly why we’re here.
Having said that, xslt is… complicated. No one maintains any of the existing implementations and building a new one would be a huge lift for what it seen as a very small user base. It’s hard to justify its place in the web.
Having said that, xslt is… complicated. No one maintains any of the existing implementations and building a new one would be a huge lift for what it seen as a very small user base. It’s hard to justify its place in the web.
What benefit do you see about getting a reference to a node without getting the tree referenced also?
November 6, 2025 at 4:59 PM
What benefit do you see about getting a reference to a node without getting the tree referenced also?
Do you have a link? What capabilities are missing?
November 6, 2025 at 4:58 PM
Do you have a link? What capabilities are missing?
Can you say more on this one. What limitations does top-layer have for you? How would a global z-index solve these?
November 6, 2025 at 2:28 PM
Can you say more on this one. What limitations does top-layer have for you? How would a global z-index solve these?
Could you say more on these?
November 6, 2025 at 1:30 PM
Could you say more on these?
It’s being discussed next Thursday IIRC!
November 6, 2025 at 1:29 PM
It’s being discussed next Thursday IIRC!
Could you tell me the kind of components you’d build without JS? I’m curious because I find it hard to imagine what could be accomplished without having to eventually resort to script.
November 6, 2025 at 1:28 PM
Could you tell me the kind of components you’d build without JS? I’m curious because I find it hard to imagine what could be accomplished without having to eventually resort to script.