Also hard not to feel like this is reaching hard to try and recreate xslt. :( It is inevitable that someone will want to template something that isn't well formed, but can combine into a well formed thing. And then you are stuck trying to find how to do it. (Or correlated entities on a page that are linked, but not on the same tree, as it were. Think "label" and "for" as an easy example in plain markup.)
If I could wave my magic wand, what we need is fewer attempts to make templates all fit in with the rube goldberg that is the standard document layout for markup. People will go through obscene lengths to recreate what judicious use of absolute positioning can achieve fairly well. Sure, you might have to do math to get things to fit, but why do we feel that is something that we have to force the machine to do again and again and again on the same data?
I was never a fan of XML, but XSLT was (is!) a killer redeeming feature of the ecosystem. And it's still widely supported in browsers! It was such a shame that XML caught on where it sucked--configuration, IPC, etc--but languished where it shined, as a markup language with an amazing transformation capability in XSLT.
I think where XSLT fell over was that it's a real DSL, and a declarative, pure, functional DSL at that. People like to talk a big game about DSLs, but inevitably they're simplistic syntactic exercises that don't actually abstract the underlying procedural semantics of popular host languages. When faced with a well-designed DSL that makes difficult tasks trivial... people can't be bothered to learn.
I didn't understand this until I spent a few years going down a rabbit hole asking questions like "why don't people like production rules?" In the case of templating people expect to make a template with a top-down organization that looks like the output, whereas XSLT really wants you build transformations from the bottom-up. On some level the bottom-up part is clear, particularly if you want to code-generate your rules (towards the end I was writing production rules that write production rules) but what is not clear is how the parts all fit together: you can't visually tell between an XSLT that builds the structure you want vs one that doesn't.
I think the most fun I ever had with XSLT was when I used an XSLT engine with user-defined procedures and had them create side effects, such as matching certain patterns in an XML document and making SQL inserts, though that was a long time ago when we were still using terrible XML parsing libraries.
I still can't wrap my head around how the neat and clean dsssl syntax, a real programming language, was replaced by an xml notation for the same: for cuntional code and a framework. because semantically, that's what xslt is: a functional language with a framework, geared at transforming xml instances.
but that syntax... and of course a much inferior and more obscure language than scheme underneath dsssl.
I also have a simple playground for XSLT: https://xsltbin.ale.sh/
But XML's syntax sucks, and so inevitably does XSLT's, because XSLT is just XML. Were it s-expressions, the syntax could suck slightly less. It was (is!) a small price to generate XSLT using XSLT, which makes XSLT very powerful and expressive if you hold it right, almost like a Lisp. This saved me a few times around year 2000 or so.
CSS and HTML have a dual relationship. You could certainly write CSS with an XML-like syntax but people would always get confused at whether they are looking at style or markup. Because HTML and CSS look completely different you never have that problem.
XSLT shares the same problem with the RDF specs coming out at the same time that it hid the production rules/logical nature of the system, had it looked more like
<someElement someAttributeattribute=$x/> -> <anotherElement>$x</anotherElement>
When I was looking for my own revolution in software engineering I saw one part of the low code/no code puzzle was that conventional tools force you to determine what order events happen which was something laymen shouldn't be bothered to do. Some counters are: spreadsheets (they figure out what order to calculate it), make (does a topological sort), dependency injection tools like Spring (writing a FactoryFactoryFactory isn't so bad, but maintaining it is a disaster when a "small" change means you have to reorder the order in which you construct everything)
There is a "freedom is slavery" problem here. People saw the semantic web as "you're going to exhaust yourself arguing with people about standards and ontologies before you even start coding" and not "if my data is properly namespace I can throw data from 10 different sources together into my RDF database and start writing queries".
XSLT pattern matching is the plain kind: here is a pattern, look for it in the input and process every match. If some part of the input document is ignored, it's just not useful; if some part of the input document is processed several times, it's perfectly well defined.
It was actually great when you got it, but the learning curve was so steep many developers couldn't use it effectively to begin with. For complex pages only certain developers could make changes or fix the bugs. Precisely because it was functional and most developers at the time really only understood imperative.
In fact, I remember the DailyWTF had a WTF about using XSLT as client-side transforms a few years later:
https://thedailywtf.com/articles/Sketchy-Skecherscom
But doing something like that was in fact so much faster than doing it in js, and when you groked it (deliberate throwback), it was so much simpler. I actually wrote a pivot table control in XSLT which completely blew away the performance of the pre-v8 javascript one. Pre-V8 javascript was so slow most developers wouldn't believe you now. A 10,000 iteration loop of even basic operations was often enough to cause IE6 to show a pop-up saying the page wasn't responding.
The pivot table in javascript would crash with just a few hundred lines of data, in XSLT it was instant with even 10,000s.
A really interesting use of XSLT on the web at the time was the WoW character viewer. You could view (and share) your character on Blizzard's website, with all their gear, skills, etc. It was blazingly fast for the time and it was all written in XSLT.
The true role of XML are grammar-based notations. These occur in two places: when a human gives data to a machine and when a machine produces data for a human. This is where XML is used despite its often mentioned shortcomings; for example, many notations to describe the user interface are based on XML. This is convenient, because user interfaces are created manually. (I am not mentioning text markup, it is well known.)
Yet XML was often used as a notation for machine-to-machine exchange. For example, the ONIX book description standard. Here data are moved between two computers, yet for some reason they have to form grammatically correct phrases according to a set of grammar rules. Computers do not need grammar. They do just fine with non-grammatical data, like a set of tables. It is way simpler for them; parsing or generating grammar, even explicit, is pure overhead for data exchange and is only necessary when data enters or leaves the computed pipeline.
So, to your examples: configuration in XML is actually fine, but IPC is not. Configuration is written by hand, IPC happens between machines. IPC specification, on the other hand, is also a good fit for XML.
That said, XML and thus XSLT has another flaw: it is way too verbose and has no good way to format it. Conciseness was an explicit no-goal but now we can say it was a mistake.
It didn't help that Microsoft dropped a stick of over-complicated standards that tried to bring RPC into XML. RPC has always been a cursed concept because between (1) trying to be intellectually coherent and (2) caring about performance RPC systems become incomprehensible and it doesn't matter if it is Sun RPC, DCOM, CORBA, "Web Services", Protocol Buffers, etc.
The fact that the "REST economy" is intellectually incoherent and could care less about performance seems to have helped it succeed. Right now I just wrote a javascript function that looks like
const get_item = async (item_id) => {...}
GET /item/{item_id}
Item getItem(String item_id)
Granted, it did seem that XML got more heavily abused than some other options for a while. I am curious if that is just a by product of when it was introduced. That or just the general proliferation of how many front end developers we have. (I hate that I am pushing that to almost be a complaint. I certainly don't mean it that way.)
XML needs another syntax that isn't so verbose. Sort of like how OWL has XML, Manchester, Functional, and Turtle syntaxes for the same data structures.
XSLT needs a Turtle-style syntax.
XML in general (the data structure, not the syntax) needs a Turtle-style syntax.
I choose to look at this a little differently.
An XML application using XSLT is so much better (faster load times, faster to write, easier to make correct) than a JavaScript application with a JSON api, that XML is basically a secret weapon.
I only care enough that it stays in browsers, but otherwise I'd prefer nobody know about it because it means less competition on things that matter (faster load times, faster to write, fewer bugs, etc). And I've got a half-baked JavaScript-based "renderer" I can finish in case some Google VP asshat goes looking for things to delete...
How do you come to this conclusion? It seems to me that what you mean is a general gripe with HTML+CSS, not with how it's generated.
And why do you bring up absolute positioning?
I hear this take on HN again and again and sure, absolute positioning has its place, and is needed for many things.
But when it's used for page/app layout, most of the time I came across this it was an absolute nightmare, falling apart at the slightest content (even text!) or screen size changes.
Even print newspaper layout can't work like this, because typography is involved, although it's probably a lot closer to what I imagine you are describing.
Maybe I'm misunderstanding you.
But when I was doing more CSS-intensive work (I still do a fair bit), developing something on a basis when someone created a layout based on absolute positioning that looked like it was "almost ready", it was a terrible time sink to try to fix it and recreating it using flex, flow et al for layout (I'm not that fond of grid outside of some scenarios, and at the time I didn't use it due to browser support) was always faster because the problems with absolute positioning as the main layout tool were basically unfixable.
Maybe there are techniques using calc() and viewport units where it makes sense, but absolute positioning is not suitable for any layout outside of completely static content and viewport dimensions, in my experience.
Basically, my assertion used to be to draw out what you have in mind on grid paper. Then start creating the elements. I don't see how that flow could land you with the 100ish divs that you wind up with on something like a single blue sky post.
Is it a panacea? No. Of course not. Can a constraint language help? I think so.
I'll add that the flex layouts seem like an improvement over what used to be there. Even if I find it rather amusing that we seem to have settled back on tables for layout. :D (I suppose it is more like some of the layout managers from Java days, truthfully.)
But, fundamentally, the problem appears to be that the drop to symbolic text is just not something that everyone does the same way. As such, it is very easy to get into a lot of trouble with the number of cooks in the kitchen, as it were.
It's not that every website uses CSS grid for layout.
Coincidentally, I took a look at the DOM+CSS of a bluesky post just a few days ago (very weird coincidence, since that was the first time I opned bluesky for months), and it did use old-school tricks like centering using CSS transforms, presumably because renders a tiny bit faster than flex centering, or avoids layout calculations when elements are added in a virtualized list.
Virtualized lists are also a good example for falling back to specifying exact pixel positions and dimensions for performance reasons, but these are usually determined with help of JS. I think the transform I saw was a translateX(-50%) one, so centering.
I totally get the canvas-like approach, but in a way the constraint-based flex layouts fall into the same line of thinking for me.
The issue with absolute positioning is the need to manually specify positions and dimensions for elements, which makes it useless unless you are working within a fixed box or only relating to the corners of one rectangle.
It is explicitly meant to remove elements from the normal layout flow so they overlap each other by default.
the web has the requirement that the 'document' look good no matter what device size/dimension, orientation, and/or capability.
In regular apps (say, a windows app), you don't have this requirement. In mobile apps, there's a standardized set of sizes. Only on web do we have both!
First of all, it's not very nice to laugh in the face of someone advocating for progress on the web platform, which benefits everyone.
Second of all, yes we do now know what good syntax for templating is, it's basically jsx (and I'm saying this as someone who's really not a fan of React). It took the whole web by storm, it's been adapted for all kinds of frameworks, and it's undeniable that all js templating systems converged towards common attributes: templates-as-expressions, composition via nesting and control flow with just javascript (instead of specific template syntax).
JSX is certainly the most popular because it's used in the most popular framework, but it has some very clear weaknesses. In particular, it has very clear semantics for a React-like, VDOM-based framework, but those semantics do not work as well for other kinds of framework.
For example, you mention control flow via ternaries/`.map`. This works great in React, where the entire template will be reevaluated every time any input changes. However, frameworks like SolidJS or Vue in Vapor mode work very differently, and typically evaluate JSX only once at component mount (or at least, as seldom as possible). To support these, these frameworks need to use either special components like `For`/`Show`, or directives like `v-if`.
There's also the issue of how to evaluate JSX. In theory, JSX is flexible, in that `<Cpt prop={expr} />` can be converted to any call of the form `h(Cpt, {prop: expr})`. Again, that's not true for SolidJS or Vapor mode Vue — in both of these frameworks, `expr` cannot be eagerly evaluated, so the traditional transforms just don't work. Both of these frameworks therefore have to include their own custom plugins for handling JSX correctly.
The author's suggestion to also use signals as a state mechanism suggests they're imagining something more along the lines of SolidJS. That's an odd choice (it's my personal go-to framework, but it's also very niche), but it also implies, as discussed, that JSX-in-the-browser wouldn't behave like JSX-in-React. From experience, that will cause problems! I've worked with React developers before who've really struggled to use SolidJS because of precisely this issue.
And it is funny, because I can already feel the ideas that would go into templating this symbolically. Characters have 6 and 20 numeric attributes. But, I can already guess most would consider it a fault to manually place either of those on the page. Yes, the sheet has a limitation on how big your name can be. No, you can't really avoid that.
JSX is what happens when you no longer have a design and a dev team. It is great at that, even. But if you have a workflow where a designer makes a template and sends it over to a dev, it no longer really helps. You are much better off making something that can pick at artifacts that are handed off.
Instead, we seem to be aiming for an idea that will have to replace learning of everyone involved.
I have no idea what substack and bluesky are, but I'll take that to suggest that someone used templating to create a mess. While that is no doubt true — someone can create a mess out of anything — would the same person have avoided the mess if the templating wasn't there? It is just ergonomics, after all, not some fundamentally different idea.
Do you have examples that are good?
printf("%d", 10);
You seem to have gone on a tangent that "good" is a general topic in my question. I meant do we have specifically good examples of templates. Surely if we know what good templating syntax is, we can share examples of it? Even if you can't describe it directly.
So, yes, I understand you are trying to call attention to my 'loosey-goosey' usage earlier. But I am saying that when I said "good", it was relative to the temporal position it found itself in.
Dodging the obvious question in favor of discussing if we can make progress... feels less than good faith.
But, as recognized earlier, others may find it doesn't hold up to today's standards. "Good" is not only not a constant, but is also subjective. Do I really have to explain the entire universe here? Man.
Which is, you guessed it, a language! Okay, yeah, I didn't dive in so deep as to provide a formal specification for the language, or whatever it is you were hoping for, but if you really want to take this to silly town, I'm going to tell you that what you saw is the only valid input for this language and only specify that, so, maybe, unless you are having fun with this comedy routine (in which case, carry on; I'm certainly still entertained!), you can read between the lines? The question asking if I need to explain the universe was rhetorical, implying that I am not going to do that.
The article fails to accept that performance and security aren’t addressed by vanity layers. This is a mistake repeated by web technologies when popular demand eventually crushes common sense, because hiring is more important than training/maintenance when the lowest levels of the work force can’t tell the difference and drives all design decisions.
If you want better performance or security you have to measure things, not wear a pretty dress and look the other way.
This proposal is a good example of how common issues with the platform are solved on top (React etc.) until we recognize them as a problem and then push them down. Polyfills are another example.
If a proposal like this succeeds, it lives a time in the sun, but then spends most of its useful life being the old thing that people are trying to work around, just like the DOM API, just like ECMA versions, just like old browsers, just like every other useful bit of tech that is part of the system but can't be touched.
Is it possible to think about entropy, extension and backcompat as primary use cases?
I want the web platform to have every possible capability that native platforms have (subject to privacy and sandboxing constraints, of course). And I want the developer experience of web developers to be incredible.
But these need to be balanced against the consequences of added complexity. And in this case, does native templating really improve developer experience? I'm not convinced the benefits outweigh the costs.
But in the process, the base functionality has been propped up another level.
Incremental updates aren't worthwhile just because of userland requirements that will always discover new gaps, use-cases and blindspots.
I dunno --- getElementById has been stable for, what, 25 years? "There's no such thing as a stable API" is something said by people unable or unwilling to create interfaces that last. It's a statement of personal resignation, not cosmic impossibility. There are tons of counterexamples.
Application needs, like other needs, are infinite. You satisfy these needs by adding new APIs, not breaking working ones.
At the same time I don't think there is actually anything that most people would consider an API that is open to public usage that has maintained that kind of stability that getElementById has, which after all is something most people would describe as a method of an API.
On the web they are. Once something is out in the open on the web, there will be people depending on this, in this exact form, forever.
That's why there are still APIs that end up in "smooshgate" because of decisions from 20 years ago: https://developer.chrome.com/blog/smooshgate
It will be near impossible to get everyone to agree on a standard template system. What the browser CAN do, however, is provide some lower level APIs on how to apply diffs to the DOM in a performant native way.
I would LOVE for something like this to exist in browsers natively:
element.applyDiff(DocumentFragment | string, { method: 'innerHTML' | 'outerHTML' })
This could apply the diff in a way that would be non-disruptive, i.e. it would keep element focus, input values, states in audio and video players, mutate attributes, etc. There are JavaScript libraries that do this like Idiomorph, but a native solution has the potential to be much more performant.
I can't even begin to imagine how much CPU and bandwidth is wasted with billions of users downloading, parsing, and executing something like React.
DOM sucks though, it's slow, it's heavyweight, it lacks transactions. We're stuck with it, and frameworks like React have to do the DOM diffing + patching thing, explicitly, in JS.
Alex Russell has written swathes of arguments about Reacts performance issues https://infrequently.org/2024/11/if-not-react-then-what/
The DOM has become much faster since React started over a decade ago, the VDOM really isn’t needed anymore even for app like experiences
React is about developer preference over user experience
(I'm still of the sort that thinks Signals are just worse Observables, so it's not a proposal I'm particularly thrilled about, but were it to be adopted Signals are easy to use in Observable contexts as well, they are just uglier half-implemented BehaviorSubjects, though maybe with a few extra lint rules to prefer Observable behaviors over Signal ones.)
by one of the people wrecklessly barging forward with half-baked specs that introduced significantly more problems than they solved, pushed a "solution" that requires 20+ new web specs to barely do all the things user-space is already doing while completely ignoring and gaslighting anyone who wasn't 100% on board with what they were doing.
Safari argued that there should be a declarative ways for this 15 years ago
<pedantic>
It's "recklessly". "reck" is a very old word meaning "to care, heed, have a mind, be concerned about"; so "reckless" means "without taking heed".
I actually thought it was directly related to "reckon" (meaning "to think or calculate"), but when I looked it up it turned out not to be the case (except much further back in the etymological tree).
</pedantic>
My brain knows it "reckless", my fingers type "wreckless". Same happens to a few other words, too.
My feeling is that they were focused on designing something that is aimed at building form controls, not the kinds of components web developers use in practice. They are designed to make browser vendors’ lives easier, not web developers. That’s often excused with “web components excel at ‘leaf‘ components” when what is actually meant is “web components are bad at everything else”.
I would expect an actually good solution that fits in with the web’s architecture to come from the direction of HTMX, not web components.
> Safari argued that there should be a declarative ways for this 15 years ago
True, but they were equally able to propose and deploy alternative solutions and mostly just went along with web components (with exceptions of course).
Safari doesn't have as many engineers (a shame) and definitely doesn't have as many people whose apparent job is just to sit on standards committees and generate specs (like Alex Russel, Justin Fangnani etc.).
They did end up proposing declarative template instantiation in 2017: https://github.com/WICG/webcomponents/blob/gh-pages/proposal... but that mostly went nowhere
It really is a shame Apple don’t invest more in WebKit and the web standards process. Although they’ve been doing a lot better over the past few years.
That proposal hasn’t just stalled, it’s been withdrawn. https://github.com/tc39/proposal-record-tuple/issues/394
It has been replaced by https://github.com/tc39/proposal-composites
I think the reason is because the DOM is a leaky abstraction and at some level I would just prefer last write wins.
I realize declarative templating is supposed to handle that, but this starts to break down really quickly when you share mutable state between components.
Second, what can you do with imperative control of the DOM that is less practical with the declarative one? I can only think of certain methods (attachShadow(), showModal()) but even then you're a 10-line component away from making it declarative.
The actual semantics for templating and data binding could just be a set of standard functions that use those syntactic feature, much like what you see in Jetpack Compose.
In fact, we could have that cross-language.
There is no value this provides over making a tagged template function and exposing it as a library. If that library is stable with ubiquitous adoption for 5-10 years then maybe there's something to talk about.
Now you have to find a way for javascript to interact with your template language.
While functions are sufficient. That doesn't look like orthogonal language design.
One reason why things like this have never happened before is because the people who need this are only barely capable of working with HTML. The DOM supports a wide variety of technologies far outside and unrelated to HTML.
There are no parts of DOM APIs that are functional. It's all 90s-era Java OOP-style.
DOM API is 90-s era OOP
setHTML() is already implemented in Chrome/Edge and Firefox so this point is a bit outdated - there is a safe alternative to innerHTML.
There is an alternative suggestion to DOM parts which might be a better bet: https://github.com/whatwg/dom/issues/736
God I love lithtml’s tagged template literals so much more than react’s JSX or Vue’s 3-in-one thing. It’s just html, in strings, in JavaScript. Lit is just a way to make custom elements easier. Man it’s gonna suck when I have to move on from my current gig and get my hands dirty with react again.
It's not. It's a custom HTML-like syntax with lots of custom and weird rules.
Just look at what happened with Web Components. It didn't take over or become the foundation of everyone's software. It just became yet another competitor [0].
I wish the standards committees would focus their efforts on improving JavaScript the language. That has a much greater and more lasting return on investment.
* always unnecessary
* always artificial
* only vanity
* only desired by insecure persons not familiar in the technology
* only qualified as bad idea but necessary because people were just going to do it anyways
So far the DOM has managed to escape this stupidity only because it is not a part of JavaScript. Java people ignorant of JavaScript desirous of features to make JavaScript feel more like Java has no bearing on the DOM, for example, because they are separate technologies managed by unrelated organizations.
None of the ergonomic reasoning mentioned in the article are qualified. Just because many people lack confidence in the technology and knowingly make poor design decisions doesn’t mean a familiar vanity layer will fix anything. Declarative comfort, for example, is not a resolution to performance and security problems just because other knowingly bad design decisions are worse. Two wrongs don’t make a right.
Furthermore the DOM already has a slow unnecessary declarative abstraction layer insecure people cannot live without called querySelectors. In other words this proposal is to React as querySelectors are to jquery, and classes are to Java. These are/were trends and trends die over time. We really should move past vanity as an alternative to an absence of training.
> Developers need to reach for a library, and thus tools like npm or a CDN, to do many basic things. This adds to the overhead of getting started. It makes simple static files and devtools less useful than they could be. There's no fundamental templating knowledge that's portable between stacks, and native DOM creation APIs like innerHTML are unsafe by default.
Remember when you could just drag an html file into your browser, and it would work? No build step, no package install, no web server, just vanilla html+css+javascript?
It would be nice to get to do that again, and the more we move things like .querySelector out of libraries like jQuery and into native browser APIs the better, imo.
That should ideally be the highest calling of frameworks like Lit and packages like Lodash - to be so good that they prove indispensable, and ultimately go native.
The answer to this is both "never gonna happen" and "you already can."
You already can ship a React app in pure JS and even import modules via ESM in the browser from CDN. Performance will suck but you can.
You'll never be able to actually have a complex web app that you can just drag into the browser. As the base API expands, so do the ambitions.
Heck we've had PHP 4 years after HTML just to fill in some blanks, people will always want more than static code.
XUL was beastly back then though.
Out of curiosity, what does that app do to convince people to jump through such hoops? Would you mind sending a link to it?
There are still 3 companies that use it (since 2008), so their employees don't have a choice really. The app does a lot, so to stop using it the companies would need to hire and migrate to 3-4 other services. I reckon SAP and the kind could do everything as well, but these companies are too small for that.
There isn't a website or anything anymore for me to show, and I haven't been involved in it for over 10 years.
https://www.mozilla.org/keymaster/gatekeeper/there.is.only.x...
Good old time.
I didn't understand this part, can anyone shed light? What is different between what's being described here and what React does with event listeners, etc?
https://react.dev/reference/react-dom/components/common#reac...
We can nitpick this point because react has had a ref API for at least 5 years now. Given a ref, all DOM API are available. For events, SyntheticEvent will refer to a native event if it exists.
The SyntheticEvent abstracts vendor discrepancy. Under the hood, react can apply some optimization too.
https://legacy.reactjs.org/docs/events.html https://react.dev/reference/react-dom/components/common#reac...
https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/...
?
The next two documents are part of a set that I made which did DOM-based templating on the back end in Java
https://ontology2.com/the-book/html5-the-official-document-l...
https://ontology2.com/the-book/source-code-transclusion-in-h...
one trouble is that systems that work at the DOM tree level are an order or two magnitudes slower than string-based templating systems. Hypothetically you could do some interesting things like hygenic macros and merge together arbitrary documents, rewriting the CSS classes and such. But by and large people find string-based templates to be good enough and don't way to pay the price for something more expensive.
So it is not yet a full, generic templating solution.
Also, this article goes on at length about how the templating needs to be "reactive" and not just "builds a DOM tree", and <slot> doesn't do that yet at all, even in the automatic behavior scenarios, it's a one time "merge".
Kicking the can along the road of the complexity of "reactive" components is a large part of how we've got the (quite basic) <template> and <slot> tags that we got, and I think why the article is still currently impractical. There needs to be more agreement on what "reactive" means. The article mentions the signals proposal, and that's one possibility that a lot of frameworks are pushing for right now, but it's still a process before browsers agree to support something like that, and something like that is a dependency before agreeing on what a "reactive" template language can be/how it would work out of the box.
More specifically, a native implementation of the "patch" function:
patch(target_dom_node, virtual_dom)
Most of the "slowness" of the DOM come from its requirement to be a 90's era Java style object hierarchy.
Don't call it "templating". Just call it "virtual dom". Everyone knows what that means.
It's not a free abstraction though.
https://github.com/WICG/webcomponents/issues/1069
My proposal only adds one native function with nothing else: no new data types, no new apis.
You'd need to spec out what that looks like. It adds one new API from the users perspective but much more from the browsers perspective.
Additionally the next generation of Frameworks do not use virtual DOM. Solid and svelte do not. Vue is moving away from it. Signals are directionally where they're all heading.
I'd love to see something that builds on the work of hyperscript and HAST. They are great models of the DOM. It would be exciting if a template language were syntax sugar.
JSX is easy to reason about because its elements are 1:1 with a single, uniform function call. That feature means JSX is always optional. Sometimes it is even more verbose or less-performant to use JSX than a hyperscript API like specifying optional properties. I think errors and call stacks are clearer than during string interpolation, but that's possibly BS.
Web components offer limited data binding and the hyperscript approach has clear control flow. The templates seem to be a source of confusion in the GH discussions.
There is still something special and pleasant about jquery because its API was a reflection of the domain. As a developer, I want to query for a node (CSS selector, xpath, etc.) to affect change and traverse to other nodes. After a beer or two I'm convinced emacs and org mode approaches are worth emulating in the web.
Great article and linked discussions. Thanks for sharing.
I get why OP likes signals. In every large enough project there is a half baked implementation of a DAG calc tree and it makes sense that a language could standardize one.
But these abstractions have a huge mental / implementation cost.
The problem, as with most engineering things is a tradeoff problem. The react model - where you just update the global state and re-render everything - is slower but easier on the brain. The signals model is faster, but so much effort.
Most apps out there don’t need to be crazy fast, and people will choose react because it’s just simpler.
But signals don’t really have anything to do with templating, do they? So why do we have to choose, could we have templating and signals as separate things?
Well OP thought about templating and realized you do need a way to tell the dom how to fit your templated node where it belongs and update it when things change.
And this is where these proposals fail. There needs to be a choice here. The API must pick a side (well technically it could allow for both, but ugh), and developers won’t ever agree which side it should go.
The big problem of UIs has always been how they update, not how they’re defined. Microsoft tried (and failed) at defining a ton of models, MVC, MVP, MVVM, and what not, all of them were painful AF. Then imgui come and say, well what if UIs didn’t have state at all. Ooh this is nice, but kind of hard on the cpu, so what do we do?
Well, perhaps one of the biggest reason for the success of web apps is in fact that the dom didn’t impose a way to bind data to its view. And so we may be doomed to framework hell.
there are multiple frameworks now that do fine-grained diffing without relying on signals, proxies, or any other reactive primitives. they basically have the top-down react model but much faster and without the weird concepts like hooks and manual/wasteful dependency arrays.
my favorite one is ivi-js: https://github.com/localvoid/ivi
it's just 8% slower than the fastest / ugliest / imperative / unmaintainable vanilla js you can eventually arrive at if all you care about is winning benchmarks.
https://krausest.github.io/js-framework-benchmark/2025/table...
One of the most useful features that could make a lot of incremental computation problems easier is "value types"[1], but unfortunately it seems that isn't going to happen anytime soon. The biggest constraint when developing an efficient UI framework with good DX is JavaScript. Also, it would be nice to have `Node.prototype.insertAfter()` :)
for perf, s/JavaScript/DOM, i think.
good DX comes from ecosystem and amount of time invested in making good tooling. JSX would be a non-starter without IDEs helping autocomplete, linting/format, syntax coloring, and webpack/babel to do the compilation.
tagged templates could reach at least the same level of DX as JSX if the community invested the resources to make that better. i'm not saying it's the right solution for a standard, but it would be way better than jsx, since tagged templates are already a standard.
and then you immediately go on to say this:
> tagged templates could reach at least the same level of DX as JSX if the community invested the resources to make that better.
So, tagged templates are also non-starters without IDEs helping autocomplete, linting/format, syntax coloring.
> i'm not saying it's the right solution for a standard, but it would be way better than jsx, since tagged templates are already a standard.
They are strings. There's no magic in tagged templates that somehow make them immediately better for some custom non-standard syntax compared to JSX.
You can't just plop a string containing lit's custom non-standard syntax into an IDE (or a browser) and expect it to just work because "it's tagged templates are standard".
For the purpose of templating in the browser there's literally no difference between standardizing a custom syntax based with JSX or tagged templates.
they're marginally better since they have a platform-defined way to deliniate static from dynamic parts. ivi _can_ work without a runtime or build-time JS parser, while JSX cannot (because jsx has to be parsed out of full blobs of js)
on the dx/ide side, sure there's not a huge amount of difference if both had the same effort invested.
My feeling is that tagged templates would actually be a worse fit in this scenario because now you would have to distinguish between "regular" tagged templates and "templating" tag templates.
Personally, I don't think that it will have any significant impact, everyone will continue using React,Vue,Svelte and it is highly unlikely that they are going to adapt this new API.
I would gladly take easier on our hardware, bandwidth and planet even if a bit harder on the developers' brains. (as a developer).
> Most apps out there don’t need to be crazy fast
I wish we recognized that we need apps to be lean.
> and people will choose react because it’s just simpler.
I think you are right, and I dislike React for this.
Microsoft used those at various times, but the only one it defined was MVVM.
MVC was Xerox PARC, MVP was Taligent.
Unless you loved IE6 of course, which was when Microsoft declared the web browser to be 'complete'.
There has been long running complaints about how many UI frameworks there are, and how often they change. It's settled down some, but I don't expect that situation to change for a long while.
Okay but just because fighting the river has become popular doesn't mean wanting to pave over the river is a good idea. It might be the logical _conclusion_ to fighting the river, but you could also just... stop fighting the river and use it the way it was intended to be used again.
It's like creating regulations which require a specific solution before that solution exists.
The solution to the "bro, just one more API, please" is to design a _transparent_ platform that is well able to "delegate" programming of new features (e.g. one implementing your favourite templating API) to third-parties in a manner that maintains their "first class citizen" status. WebAssembly was a move in that direction because it's a generic platform that in part supercedes and otherwise supplants the mess that JavaScript has to manage bridging the originally "kiddie script" application software domain, with the native functionality the browser may be encapsulating (also for performance).
Case in point: FFMpeg may be compiled to a WebAssembly module, which gives you arbitrary video/audio encoding/decoding -- pending correct design of bit-blitting so the decoded output can be efficiently transferred to the screen/speakers (which, for much of the reasons I am trying to outline, _is_ the bottleneck of the entire solution).
We need more of the same kind of thinking. Stop begging Web browser vendors / w3C / WHATWG for more features that are just lipstick on a pig -- sit down, think about what kind of feature(s) would allow the Web platform to finally escape the death spiral it's been in since its inception -- albeit one with a large enough radius it's meant to never actually resemble a spiral.
I don't know if I am making myself clear here, but in much simpler terms: why should there be another piece of code that caters to "most" (because you happen to be a FP/React zealot, for better or for worse) when these people can ostensibly write such templating system themselves, publish it on e.g. NPM and/or pull it and use it from there?
In fact, you could call JSX a "Dynamic Templating System" and that's a reasonable summary of what it is (in addition to other things of course).
There might be some ways that React itself could, internally, notice the special cases and special times where it _could_ be slightly more performant from using a lower level of templating, as an optimization, but I'd certainly prefer that to be abstracted away and buried deep inside React, rather than ever having to think about it myself, at the JSX layer.
Someone can let me know if React is already leveraging this for browsers that support it, I didn't research that.
1. Making it possible to do something like <template src="..."> and being able to load them from an external source
2. Making them "dynamic"
3 (and the most controversial one) that all CSS, HTML and Javascript (if you don't hate it) could be written natively like QML - one syntax to rule them all.
I've done that, requires no build step/npm/whatever. It was posted on HN for discussion a week ago: https://github.com/lelanthran/ZjsComponent
#3 is a tricky one syntactically because HTML needs to be used by mere mortals and JS is a programming language used by us gods, so unifying all three would br tricky, but again I agree with you that would be awesome. Maybe some flavor of LISP would be both "powerful like a language" and "easy like a document".
I don’t see any reason a browser level “here’s new DOM you diff and apply it” couldn't exist and be a huge win for React and other libraries, with React so much more popular than every other framework combined, and that being a pretty low level API, it makes sense to start there.
Building the overly abstracted thing first is a mistake web API authors have made too many times (see web components).
> its API was a reflection of the domain. As a developer, I want to query for a node (CSS selector, xpath, etc.) to affect change and traverse to other nodes
That's what I miss about it.
Genuinely asking, I have no clue what's being alluded to without being clearly mentioned in this thread.
Array.from adds friction. The need to wrap querySelector in null checks adds friction. The fact that they are not composable in any way, shape, or form, with any DOM methods (and that DOM methods are not composable) adds friction.
jQuery was the fore-runner of fluid interface design. Nothing in the DOM before, then, or since ever thought about it. Even the new APIs are all the same 90s Java-style method calls with awkward conversions and workarounds to do anything useful.
That's why sites like "You don't need jQuery" read like bad parody: https://youmightnotneedjquery.com
E.g. what happens when it's not just one element?
$(el).addClass(className);
// vs.
el.classList.add(className);
$(selector).filter(filterFn);
// vs.
[...document.querySelectorAll(selector)].filter(filterFn);
---
[1] This is the entirety of methods exposed on NodeList https://developer.mozilla.org/en-US/docs/Web/API/NodeList
Instance properties
- length
Instance methods
- entries() // returns an iterator
- forEach()
- item()
- keys()
- values()
ES2025 added map, filter, flatMap, reduce, forEach, and several other methods to all iterators (including NodeList directly, I believe, but definitely its entries(), keys(), values(), if not) [1]. It'll be a year or two at current pace before that is "widely accepted baseline" in browsers, but it's great progress on these sorts of complaints.
[1] https://2ality.com/2025/06/ecmascript-2025.html#iterator-hel...
I listed all public methods and properties of NodeList. It does have a forEach, so there's not much need for `for of`
As for iterator methods, I completely forgot about that :) Yeah, you can/will be able to use them on .entries()
But to me, these are minor inconveniences and habits. A thin wrapper can get you there easily if you care enough. I personally dislike this array/element confusion that jQuery adds.
But that's more and more friction. A wrapper here, a wrapper there, and if here, a try/catch there. At one point you are reinventing significant chunks of jQuery
- a plugin system
- its custom selector parser (because it had it before querySelector and is not totally compatible with it)
- its abstraction to iron out browser differences (old IE vs standard, notably) that's not relevant anymore
- its own custom (DOM) event management
- its implementation of methods that are now standard (ajax & trim for instance)
I recognize that the DOM API could be better, and comes with friction. Back then, ironing out the browser differences and its selector feature were killer features of jQuery. Today, I do not think the quirks of the DOM API warrant importing a library like jQuery.
¹ but indeed, very lightweight compared to modern frameworks like Angular, React and Vue, with all the ecosystem which comes with each of them (react-router, redux, etc).