* Add failing JSDoc property svelte-migrate conversion tests
* Add further test case and remove default value in JSDoc output
* Look for inlined JSDoc comments after a hyphen
* Add changeset
* move parent property onto Signal
* don't self-invalidate when updating a source create inside current reaction
* lazily create deep state with parent reaction
* no need to push_derived_source with mutable_state, as it never coexists with $.derived
* reduce indirection
* remove state_unsafe_local_read error
* changeset
* tests
* fix test
* inelegant fix
* remove arg
* tweak
* some progress
* more
* tidy up
* parent -> p
* tmp
* alternative approach
* tidy up
* reduce diff size
* more
* update comment
Fixes#14072
`:has()` was matching only against descendants or siblings, but not sibling's descendants. This makes the logic be able to go forward or backwards, simplifying a lot of cases along the way.
For instance, specifying a tree-structural pseudo-class to
`::view-transition-new` should still constitute a valid global-like
selector.
Fixes#15312
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* ix: ensure transient writes to tracked parent effects works as expected
* lint
* format test
* tweak changeset
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: on teardown, use the last known value for the signal before the se
* fix: on teardown, use the last known value for the signal before the se
* fix: on teardown, use the last known value for the signal before the se
* fix: on teardown, use the last known value for the signal before the se
* fix: on teardown, use the last known value for the signal before the se
* Update packages/svelte/src/internal/client/reactivity/props.js
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* Update packages/svelte/src/internal/client/reactivity/props.js
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* Update packages/svelte/src/internal/client/reactivity/props.js
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* lint
* lint
* lint
* Update .changeset/sharp-elephants-invite.md
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix null and warning for local handlers
* test
* changeset
* treat `let handler = () => {...}` the same as `function handler() {...}`
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* add style attribute when needed
* set_style()
* to_style()
* remove `style=""`
* use cssTest for perfs
* base test
* test
* changeset
* revert dom.style.cssText
* format name
* use style.cssText + adapt test
* Apply suggestions from code review
suggestions from dummdidumm
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* fix priority
* lint
* yawn
* update test
* we can simplify some stuff now
* simplify
* more
* simplify some more
* more
* more
* more
* more
* more
* remove continue
* tweak
* tweak
* tweak
* skip hash argument where possible
* tweak
* tweak
* tweak
* tweak
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* style must be set via set_attribute
* test
* changeset
* add empty string and null in test
* explanatory comment
* this is now redundant, set_attribute takes care of it
* drive-by
* tweak changeset
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: correctly set `is_updating` before flushing root effects
* rename for consistency with update_effect
* use var
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
This ensures someone like Astro can render multiple islands and the unique ids produced in them are able to be deduplicated
---------
Co-authored-by: Hugos68 <hugokorteapple@gmail.com>
Co-authored-by: paoloricciuti <ricciutipaolo@gmail.com>
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
* fix: catch error on @const tag in svelte:boundary
* changeset
* edit changeset
* test
* fix
* rollback
* simpler way to do that
* rollback
* roolback again
---------
Co-authored-by: paoloricciuti <ricciutipaolo@gmail.com>
* this appears to be unnecessary
* DRY out
* this doesnt appear to do anything useful
* simplify
* remove unused if block
* simplify, make non-recursive
* unused
* DRY
* simplify
* tidy up
* simplify
* changeset
* unused
* Revert "changeset"
This reverts commit 946e00dcf7.
* make flush_sync non-recursive
* fix flushSync types
* fix
* unused
* simplify
* tidy up
* tidy up
* present unnecessary microtasks, avoid flushing if no function provided
* simplify
* fix: `muted` reactive without `bind` and select/autofocus attributes working with function calls
Co-authored-by: Joel Howse <joel@helicalco.com>
* chore: add comments
---------
Co-authored-by: Joel Howse <joel@helicalco.com>
- `$derived` can contain `$state` declarations so we cannot ignore them, so this reverts #14533
- instead, we add equality checks to not do this expensive work unnecessarily
- this also adds a render effect similar to the class ownership addition when it detects a getter on a POJO during ownership addition
fixes#15164
* fix: when an unowned derived is tracked again, remove unowned flag
* fix: when an unowned derived is tracked again, remove unowned flag
* test case
* test case
* cleanup unused logic
* simplify
* simplify
* simplify
* simplify
* add test
* fix other issue
* back to original plan
* Apply suggestions from code review
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
Avoid going through the `element.value = element.__value = newValue` condition because `__value` is actually how Lit stores the current value on the element, and messing with that would break things: Lit would think the value hasn't changed (because `__value` is already set to the new value by us) and doesn't fire an update.
fixes#15194
* fix: ensure tracking returns true, even if in unowned
* fix: ensure tracking returns true, even if in unowned
* Update packages/svelte/tests/runtime-runes/samples/effect-tracking-unowned/main.svelte
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
`slot` is treated as a regular prop if it is not used directly inside another component, but we were running validations on such regular props that should only be run on real slots.
Fixes#15125
Fixes#14910
The issue occurs only when :has() targets at a component's root element and because include_self is false.
I came to the conclusion that this is the same case as :root:has(.scoped).
Fixes#15027 and fixes#14995
When the target element is inside a block or component, get_following_sibling_elements was looking only for sibling elements after the block/component. This PR fixes it by starting the search from the begging of the block/component and skipping nodes that go before the target element.
* fix: ensure signal write invalidation within effects is persistent
* fix: ensure signal write invalidation within effects is persistent
* fix: ensure signal write invalidation within effects is persistent
* address feedback
* fix: Make Tween duration 0 set current to target immediately
* Run prettier on test file
* tweak
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
... if it's not followed by another newline, according to the spec
Fixes#14767
---------
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
* fix: ensure unowned deriveds correctly get re-linked to the graph
* fix: ensure unowned deriveds correctly get re-linked to the graph
* fix: ensure unowned deriveds correctly get re-linked to the graph
* add test
* add test
* cleaner apporach
* cleaner apporach
* cleaner apporach
* cleaner apporach
* handle invalid expression inside each key
* handle invalid expression inside each expression
* handle invalid expression inside await block
* handle "in the middle of typing" components with starting lowercase and dot at the end
* changeset
* fix: make `defaultValue` work with spread
* chore: apply suggestions from review
* tweak
* only stash defaults when relevant
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: avoid mutation validation for invalidate_inner_signals
* add test
* Update packages/svelte/src/internal/client/runtime.js
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
fixes#14532
This removes the `reactive_declaration_non_reactive_property` warning altogether. The first version caused many false positives at compile time. The refined runtime version (introduced in #14192) was hoped to fix this, but it turns out we now get loads of false positives at runtime.
fixes#14391
In #13337 the "when to set as a property" logic for custom elements was adjusted. A bug was introduced during this, and it consists of several parts, of which the latter I'm not sure what's the best solution, hence opening this to discuss.
The problem is that during set_custom_element_data, get_setters is the only check done to differentiate between setting the value as a prop (has a setter) or as an attribute (doesn't have a setter).
The solution is to take into account whether or not the custom element is already registered, and defer getting (and caching) its setters until then. Instead, fall back to a "an object is always set as a prop" heuristic.
* chore: provide links to documentation for errors/warnings
closes#11305
* changeset
* fix most of the tests
* fix mutations messing with nodes between runs
* more concise
* more test fixes
* last one
When the `value` or `checked` attribute of an input or the contents of a textarea were static, setting the `defaulValue/defaultChecked` property caused the latter to take precedence over the former. This is due to how we transform the code: If the value is static, we put it onto
* tests
* typings
* implement for defaultValue/defaultChecked on inputs
* docs (draft)
* selected
* fix test
* remove
* tweak
* changeset
* untrack reads, they could be inside an effect
* Apply suggestions from code review
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* handle select reset case
* handle reset case specifically: use different props/queries in that case
* enhance test
* fix
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: better error messages for invalid HTML trees
closes#13331
* fix test
* more concise
* tweak
* tweak messages
* adjust tests
* tweak message slightly, so it doesn't sound like the bad element is the one we're currently encountering
* put locations in generated message
* tidy up
* consistency
* fix
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* add failing test for #14268
* simplify
* proxify values when using ||=, &&= and ??= assignment operators
* proxify values assigned to private state fields
* changeset
* fix
* fix
* add warning
* update test
* turn `reactive_declaration_non_reactive_property` into a runtime warning
* ignore warning
* Update packages/svelte/src/internal/client/reactivity/effects.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* Update packages/svelte/src/internal/client/runtime.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* fix
* test
* changeset
* Update .changeset/witty-turtles-bake.md
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* add some details
* check
* regenerate
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* Fix `$.update` and `$.update_pre` for bigints
* resolve conflicts
* fix some things
* fix thing i definitely didn't just break
* hopefully this will fix it
* fix formatting
* simplify
* style consistency
* simplify
* changeset
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: upgrade to esm-env 1.2.1 to fix issues with non-Vite setups
* fix (albeit with annoying warnings)
* bundle with conditions to silence noise
* production is probably better
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
fixes#14466
The logic introduced in #14395 was flawed - not every text or expression outside the template is the child of an attribute. This turns it around: We know that every child of a fragment is inside the template, so we ignore all text/expression tags that are not child of a fragment
Fixes#14399
Add a mechanism to connect render tags to snippets to know where to walk when coming across render tags
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
fixes#14439
This bug was introduced in #13642 because setting the input to null means the equality check ("is the input different") fails if you set the value to null
Also fixes#14441 - this bug was present for a long time, and the reason is the same as for the other bug: The equality check always returns "yes this is the same" if the value is undefined initially. The fix is similar; we need to initialize the input to something that can never be equal to whatever value is passed
This reverts #13255 / #13158, and helps with the validation error part of #14120
If you would have a component like this...
```svelte
<td>hi there</td>
```
...and then render it on the server via our `render` function like this:
```js
const result = render(Main);
```
...then right now you get an error saying that `td` is not valid at this position. But that doesn't seem right, because we should give people the benefit of the doubt: It may very well be that someone renders such a component and then correctly puts it into a `tr` tag themselves on the server (another example is rendering a full html document like in #14120).
All the other validation where there's a known parent (X not valid inside Y) is untouched by this.
Doing this as cleanup prior to tackling #13331
* fix: always use set for private identifiers
* we can simplify this further - no need to check the value was transformed, since the outcome of not returning immediately is the same but with extra steps
* add explanatory note
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
through #14363 I noticed our `export default` validation wasn't quite right:
- missed checking for derived/state exports
- the "cannot have a default export" error was only thrown if you did `export default` from the instance script, but it shouldn't matter from which component part you export it; it's never ok
* fix: ensure internal cloning can work circular values
* better fixc
* 'original' feels slightly clearer than 'json_instance'
* use an optional parameter, so we can omit it in most cases
* Update packages/svelte/src/internal/shared/clone.js
Co-authored-by: Rich Harris <rich.harris@vercel.com>
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
The detection whether or not props are writable is buggy; it doesn't take into account the case when instantiating components via `mount` or legacy-`new`
Fixes#14161
This posed an interesting question: What to do about (non-)`bind:`able properties? The answer I arrived on was: Treat it as if everything that _can_ be bound _is_ treated as bound, and everything else as readonly. In other words, if you're reassigning a prop, it will diverge from the passed in props if it's not bindable or not set in the parent, otherwise it will mutate the passed in props. I think that makes the most sense, given that you can't control this behavior from the outside.
* fix: avoid marking subtree as dynamic for inlined attributes
* fix: i'm a silly goose 🪿
* chore: refactor `is_inlinable_expression` to accept the attribute
* feat: inline dom expression too
* fix: special case literals with `"` in it and fix standalone case
* chore: simpler check first
Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
* typo
* add more stuff to snapshot test
* simplify/speedup by doing the work once, during analysis
* simplify
* simplify - no reason these cases should prevent inlining
* return template
* name is incorrect
* name is incorrect
* fix escaping
* no longer necessary
* remove obsolete description
* better concatenation
* fix test
* do the work at runtime
* fix another thing
* tidy
* tidy up
* simplify
* simplify
* fix
* note to self
* another
* simplify
* more accurate name
* simplify
* simplify
* explain what is happening
* tidy up
* simplify
* better inlining
* update test
* colocate some code
* better inlining
* use attribute metadata
* Update packages/svelte/src/compiler/phases/2-analyze/visitors/Identifier.js
Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
* Apply suggestions from code review
---------
Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <rich.harris@vercel.com>
We previously marked all `:root` selectors as global-like, which excempted them from further analysis. This causes problems:
- things like `:not(...)` are never visited and therefore never marked as used -> we gotta do that directly when coming across this
- `:has(...)` was never visited, too. Just marking it as used is not enough though, because we might need to scope its contents
Therefore the logic is enhanced to account for these special cases. Fixes#14118
While fixing this I cleaned up some inconsistencies in what we mark as global. This simplified code and fixed some adjacent bugs, which conindicentally also fixes#14189
* fix: consider static attributes that are inlined in the template
* fix: use `is_inlinable_expression`
* fix: move check for inlinable expression as last
* fix: simplify and correct
* chore: accept single node in `is_inlinable_expression`
* chore: update comment
* chore: add snapshots for non static nodes
Previously, we were applying an explicit nesting selector to the start of a relative selector chain only when starting the traversal. Prepending the selector is important because it ensures we traverse upwards to the parent rule when the current selectors all matched and there's still more to do. But we forgot to do the prepend for parent rules, which meant that if we were nested two levels deep, we would stop too early. This fix ensures we prepend in that case, too.
Fixes#14178
fixes#14168
This reverts the whole "selectors inside `:not` are scoped" logic. Scoping is done so that styles don't bleed. But within `:not`,everything is reversed, which means scoping the selectors now means they are more likely to bleed. That is the opposite of what we want to achieve, therefore we should just leave those selectors alone.
The exception are `:not` selectors with descendant selectors, as that means "look up the tree" and we need to scope all ancestor elements in that case.
If the contents of a `:not` selector don't match, then it's actually a match for `:not` because it's inverted. Therefore, we need to scope such elements. We're also making sure that contents of `:not` that never match actually count as a used (because the result is negated), and as such the contents of `:not` that always match are actually marked as unused.
Fixes#13974
`<a>` tags are valid in both the SVG and HTML namespace. If there's no parent, we therefore have to look downwards to see if it's the parent of a SVG or HTML element.
fixes#7807fixes#13793
We didn't account for the `$props` rune being writtin in a way that makes some props unknown, and they would only be visible through the `customElement.props` definition. This changes the iteration to account for that and also adds a note to the documentation that you need to list out the properties explicitly.
fixes#13785
Fixes#13848.
When we set custom element attributes/props, we should be doing so without the current effect/reaction active. Otherwise, the custom element lifecycle might attach effects/dependencies to the wrong reaction and all manner of things can incorrectly occur
---------
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
We originally didn't extend from `Error` anymore because its fields are of no real value to us, and has problems with serialization in a worker context.
Turns out this was a mistake, because various build tools rely on errors being thrown as something that extends Error, else they try to wrap it in their own error.
We therefore revert that change while still trying to preserve most of the advantages of not extending `Error`, namely nuking the useless stack trace and making sure the message is enumerable.
immediately after merging #14006 I noticed there's a bug when pruning one selector in a list of several; this fixes it
no changeset because not released yet
When CSS is externalized we rightfully rely on the following tooling chain to properly minify CSS. When we inject the CSS however, that tooling won't be able to do that, so we gotta do it ourselves.
This PR brings back most of that logic that existed in Svelte 4. Fixes#13716
Only count down after timeout, else we would reach 0 before our own render effect reruns, but reach 1 again when the tick callback of the prior teardown runs. That would mean we re-subcribe unnecessarily and create a memory leak because the old subscription is never cleaned up.
* fix: add empty stack to `CompileDiagnostic` to show error on build
* chore: generate types
* chore: make stack optional to make TS happy
* chore: generate types -.-"
* chore: add comment
* chore: highlight swallowed errors from await blocks in DEV
* chore: highlight swallowed errors from await blocks in DEV
* chore: highlight swallowed errors from await blocks in DEV
* lint
* feedback
* feedback
* add test
* Update packages/svelte/tests/runtime-runes/samples/await-no-catch-error/main.svelte
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* fix: ensure SVG element attributes have case preserved
* fix: ensure SVG element attributes have case preserved
* fix: ensure SVG element attributes have case preserved
* fix: enable bound store props in runes mode components
* add some JSDoc, since it could be a headscratcher for future us
* make it clear that this is specifically about bindings
* skip intermediate value
* tweak other names too
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: internally wrap store subscribe in untrack
* lint
* Update packages/svelte/src/store/utils.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
Closes#13493.
This PR allows the usage of getContext() inside $derived runes. Previously, you could use it, but only on init and not updates – and this inconsistency was unnecessary. We can make it work just like we do in other places.
I resisted this previously because it felt a bit wasteful, but I now think that there's really no way around this: Instead of only going upwards the tree while matching, for `:has` we go _down_ the tree to see what matches. More specifically, we're collecting the children of the current element and then check if one of those does match the selectors inside `:has`.
This makes the way the code works easier to reason about and also removes some boolean tracking we had to add for the previous approach.
Fixes#13779
Fixes#13717
There are two parts to this:
1. the parent selectors weren't passed along for the check inside `:has`, which in case of a leading combinator would mean it would always count as unused
2. In case if a selector like `x > y:has(z)`, the prior logic would correctly determine that for element `z` there's a match for the `:has` selector, by first checking its contents and then walking up the tree. But after it did that, it would try to walk up the tree once more, which is a) wasteful b) buggy because the tree walking mechanism would no longer be adjusted for the `:has` special case, resulting in false negatives. To fix that, the `:has` will return a new value from the function, signaling that it already fully checked the upper selectors, and so the function calling it will skip doing that.
- detect store mutations and not use `$derived` in that case, fixes#13723
- better detect `let x` that can be folded into `$derived`, fixes#13727
---------
Co-authored-by: ComputerGuy <63362464+Ocean-OS@users.noreply.github.com>
- correctly assign children snippet to default slot, fixes#13067
- allow `svelte:fragment` without `let:` directives to be rendered by `@render children()`, fixes#13066
* fix: moving deriveds during migration deletes part of the inserted code
* fix: use update instead of remove
* Update .changeset/cool-apes-confess.md
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
The other part of #13395
This implements scoping for selectors inside `:not(...)`. The approach is almot the same as for `:is/where(...)`.
This is a breaking change because people could've used `:not(.unknown)` with `.unknown` not appearing in the HTML, and so they need to do `:not(:global(.unknown))` instead.
While implementing it I also discovered a few bugs, which are fixed in this PR:
- `foo :is(bar baz)` wasn't properly handled. This selector can mean `foo bar baz` but it can also mean `bar foo baz` (super weird, but it is what it is). Since our current algorithm isn't suited for handling this, we just assume it matches and scope it. Worst case is we missed a prune
- `bar` in `:global(foo):is(bar)` was always marked as unused, even if it matched
Paolo Ricciuti
Rich Harris
Robert Gieseke
adiGuba
7nik
Éric NICOLAS
Dominic Gannaway
Simon H
Tom Adler
Ben McCann
Alexis
Ranjan Purbey
tomoam
Caique Torres
Elliott Johnson
Trevor N. Suarez
Edoardo Cavazza
Theodore Brown
ComputerGuy
Matthijs
Andrey Sitnik
Taku Fukada
Yang Pan
Santiago Cézar
Torsten Dittmann
Gonzalo Ruiz
Conduitry
FoHoOV
Ahmad S.