Provides a migration function, exported as `migrate` from `svelte/compiler`, which tries its best to automatically migrate towards runes, render tags (instead of slots) and event attributes (instead of event handlers)
The preview REPL was updated with a migrate button so people can try it out in the playground.
closes#9239
* feat: only inject push/init/pop when necessary - closes#11297
* regenerate
* differentiate between safe/unsafe
* only inject $$props when necessary
* more
* fix
* simplify
* handle store subscriptions
* fix: mark `accessors` and `immutable` as deprecated
* add warnings for deprecated <svelte:options> attributes
* disable accessors in runes mode
* update tests
* tidy up
* the hell?
* regenerate types
* if I would get a dollar for every windows bug I fix I would be a millionaire by now
* return instance _and_ props in runes mode, move flushSync into shared code, don't set accessors in runes mode
* goddammit
* note breaking change
* fix
* regenerate messages
* Revert "return instance _and_ props in runes mode, move flushSync into shared code, don't set accessors in runes mode"
This reverts commit a47827e57d.
* pass instance to tests
---------
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
This is a typings PR and the companion PR to sveltejs/language-tools#2336
It introduces two new types:
- Binding: Marks a property as being bound (i.e. you must do bind:x)
- Bindable: Marks a property as being able to be bound (i.e. you can do bind:x)
Language tools then uses this generate code accordingly which then generates type errors.
All the other type gymnastics are there to ensure that you don't interact with these bindable types when using mount or hydrate or ComponentProps<MyComponent>, i.e. these two types should be mostly opaque for day-to-day users.
For backwards-compatibility, all properties are automatically wrapped with Bindable, which means existing type definition files will continue to work from a types perspective. Language tools opts into strict bindability by providing its own constructor definition for all generated classes in runes mode which omits the "wrap everything with bindable" behavior.
* breaking: disallow binding to component exports in runes mode
Svelte 4 allowed you to have `export const foo = ..` in component A and then do `<A bind:foo />`. This is confusing because it's not clear whether the binding is for a property or an export, and we have to sanitize rest props from the export bindings.
This PR therefore introduces a breaking change in runes mode: You cannot bind to these exports anymore. Instead use `<A bind:this={a} />` and then do `a.foo` - makes things easier to reason about.
* Update sites/svelte-5-preview/src/routes/docs/content/03-appendix/02-breaking-changes.md
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* tweak messages
* fix tests
* use component.name
* oops
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: run event attributes after binding event listeners
By running the event listener logic inside an effect on the first run we guarantee that they're attached after binding listeners. Fixes#11138.
* give arrow functions stable id, better code gen
* optimise normal function expressions too (rare but valid!)
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* Revert "fix: take outroing elements out of the flow when animating siblings (#11208)"
This reverts commit c44234dc2f.
* fix: measure elements before taking siblings out of the flow
* lint
* add changeset back
* changeset
* fix: fall back to component namespace when not statically determinable
In #10006 we added more elaborate mechanisms to determine which namespace a given element is in. For `<svelte:element>` we added a "can't know at compile time" case and introduced a limited heuristic into the runtime.
This doesn't work for a few reasons:
- we're checking the parent's namespace to determine the current namespace, but the element itself could be the one that _changes_ the namespace
- as mentioned in the previous comment already, on the first render we can't do any parent analysis
- it does not take into account the static component namespace
The last point is the crucial one: In Svelte 4, we're falling back to the component namespace if we can't know statically - e.g. if someone added `<svelte:options namespace="svg">` then `<svelte:element>` should fall back to that namespace instead.
We were not doing that up until now, which introduced a regression. Fixing this also means getting rid of the (flawed) "can't know statically" heuristic.
Fixes#10858, though for a complete solution we likely need some way to tell `<svelte:element>` the namespace at runtime through a special attribute. Maybe we can use `xmlns` for that like we do in the static case
* support dynamic svelte:element namespace through xmlns attribute
* fix
* feat: allow $inspect reactivity map, set, date
* feat: inspect map without adding new data source
* feat: add inspect
* feat: define inspect on dev mode only
* fix: lint error
Ownership was not widened when assigning a sub state to a different top level state. The set of owners for the state was zero because the owner was on the original parent, but that one was reset to null because it's now the top level of a different state. That meant that there was no owner but also no parent to check for the owner, which is an invalid combination resulting in a nullpointer (and also potentially false positive warnings in other situations).
fixes#11204
* breaking: warn/error on old syntax in runes mode
- warn on slots and event handlers in runes mode
- error on `<slot>` + `{@render ...}` tag usage in same component
closes#9416
* render tag + slot could occur in legacy mode as well, error there, too
* rename metadata.o to metadata.owners, tweak check_ownership implementation
* track parent relationships
* update
* changeset
* adjust test to reflect new semantics
* prevent component using bind: with object it does not own
* failing test
* fix#11060
* add explanatory comment
* don't accidentally narrow global ownership, fix has_owner method
* widen ownership if assigning different state proxies to one another
* don't set owners to null when parent exists
* fix
* only recurse into POJOs
* handle cycles on context
---------
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
Co-authored-by: Dominic Gannaway <dg@domgan.com>
- don't call fallback values eagerly, only when it's actually needed. Avoids potential unwanted side effects
- use derived_safe_equals to memoize results of destructured snippet/each context values with default values to ensure they're only recalculated when their dependencies change. Avoids unstable default values getting called multiple times yielding different results. Use derived_safe_equals to ensure new values are always set, even when mutated. fixes#11143
* added raw elements set
* added test
* added changeset
* moved raw text elements to constands and made array
* moved to correct constants
* fix test
* fix constants formatting
* feat: hot module reloading support for Svelte 5
* fix lockfile
* tweaks
* types
* lint
* lint
* tweaks
* add hmr flag
* tweak
* tweaks
* move HMR logic into its own module
* simplify
* tidy up types
* fix test
* lint
* need some indirection here or references break
* prevent transitions during HMR update
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: loosen proxy signal creation heuristics
* tighten up test
* update comment
* no need to create a source outside an effect here, because it can't result in a reference
* preserve reference to props.$$events
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
- take into account that template could consist of a single script tag, for which querySelectorAll('script') would yield false negatives
- add test to ensure that we don't execute script tags inside `@html` tags next to static script tags
fixes#11082
Previously, ownership widening/removal was only done if the immediate object that was encountered was state. This isn't always the case. It also didn't take into account classes with state on it (which turn into getters). This change takes both these cases into account and now always traverses the given object deeply.
fixes#11060
- don't throw a dev time error when binding to an export (fixes#11008)
- remove bindings that are for component exports
- throw an error when using a component export with the same name as a property
* fix: hydrate HTML with surrounding whitespace
* add test
* fix a few more short comments
* tidy up
* avoid magic strings
* avoid magic strings
* fix
* oops
* use server-rendered HTML as hydration test starting point
* update tests
* remove _before.html files
* remove _before_head.html files
* override output with _expected.html
* expected output for binding-input case
* remove unused files
* fix
* changeset
* feat: take form resets into account for two way bindings
When resetting a form, the value of the inputs within it get out of sync with the bound value of those inputs. This PR introduces a reset listener on the parent form to reset the value in that case
closes#2659
* slightly different approach
* tweaks, test
* this is a breaking change, strictly speaking
* bind:files
* use capture phase
* tweak wording
* use promise, explain
* breaking: apply fallback value every time in runes mode
closes#9948
* apply fallback value in setter
* encapsulate fallback logic
* we should keep this logic out of b.set, since it's very specific to accessors
* oops
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: improve element class attribute behaviour
* Update packages/svelte/src/internal/client/dom/elements/class.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* follow up to 10846
* lint
* simplify
* don't update value
* rework logic, rely more on mutation observer, fix bug related to select multiple
* fix lazy select options bug
---------
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
Co-authored-by: brunnerh <brunnerh@users.noreply.github.com>
* feat: add support for webkitdirectory DOM boolean attribute
* add to types
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* feat: adds reactive Map class to svelte/reactivity
* add docs
* add docs
* add test case
* types
* make reactive set better
* address feedback
* fix typo
* more efficient initialisation
* this is incorrect, it would fail if given a map for example
* increase consistency (with e.g. proxy.js)
* tidy up
* Revert "more efficient initialisation"
This reverts commit 29d4a8078b.
* efficient initialization, without bugs this time
* convention
* delete make_iterable
* update changeset
* efficient initialization
* avoid generator functions
* Update sites/svelte-5-preview/src/routes/docs/content/01-api/02-runes.md
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
Co-authored-by: Rich Harris <richard.a.harris@gmail.com>
* fix: invalidate store when mutated inside each block
fixes#10771
* Update packages/svelte/src/compiler/phases/3-transform/client/visitors/template.js
We can simplify pre effects by not doing the flush logic at all now. Instead we can move the flushing logic to the only place its needed – for beforeUpdate
* feat: add reactive Set class to svelte/reactivity
* add some type safety
* simplify, read entries lazily
* failing unit test
* fix deletions
* minor tweaks
* work around effect ordering bug
* simplify, make entries lazy
* small tweak
* use var, minor tweaks
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* Docs: Change all initizlisation to initialisation
* Reverted to "initialization" when asserting errors with specific message
* Reverted to "initialization" where related to specific error message
- run reactive statements only once per tick, even when they have indirect cyclic dependencies. Made possible by adding an array to the component context, which is filled with identifiers of the reactive statements, and which is cleared after all have run. Reactive statements rerunning before that will bail early if they detect they're still in the list
- part of the solution is to run all reactive statements, and then all render effects, which also fixes#10597
* fix: improve bind:this support around proxyied state
* fix: improve bind:this support around proxyied state
* fix: improve bind:this support around proxyied state
* fix: treat snippets like normal components when inferring namespace
* n
* simplify
* better desc
* slight adjustment
* feedback
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* feedback
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* skip html tag
* test
* changeset name
* cleanup
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
- widen ownership when getContext is called, part of #10649
- widen ownership when assigning one proxy to another
- skip first 4 stack trace entries and bail if first after that is not a module, hinting at a mutation encapsulated in a .svelte.js file; part of #10649
* feat: improve ssr html mismatch validation
* update types
* Update packages/svelte/src/internal/server/index.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* Update packages/svelte/src/compiler/validate-options.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* feedback
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
This has a lot of overhead for large lists, and we can at least diminish in the "no state proxy" case by applying a sensible heuristic:
- If the value passed is a state proxy, read it
- If not, and if the value is an array, then bail because an array of state proxies is highly unlikely
- Traverse the first level of properties of the object and look if these are state, if not bail. State proxies nested further down are highly unlikely, too
part of #10637
It's a warning because even when typing it out and knowing what to do you'll always be in a state where the validation kicks in, and it would be too distracting to always see a compiler error during that short time frame.
closes#10374
People could've done bind:this and called instance methods on the instance - a rare case, but not impossible. This shims $set and $on when in legacy compat mode. $destroy is never shimmed because you shouldn't manually destroy a component, ever, and there's no way to make that work in the new world.
closes#10420
fixes#10271
The prior fix in #10307 wasn't quite right, because JSDom has a difference between window and event.composedPath which sounds like a bug, and a workaround in that PR made it so the test did pass when the bug still occurred. That's also why there isn't a test here, because we can't properly test this using JSDom.
* fix: correctly handle proxied signal writes before reads
* managed
* Update packages/svelte/src/internal/client/proxy.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
fixes#10511
We used the animation frame dance previously where the `$effect` timing was slightly different and did not wait on its children before rerunning. With that changed now everything false into place nicely.
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
The previous `bind_this` implementation had flaws around timing and didn't work correctly when the binding wasn't `$state` in runes mode.
This PR reimplements `bind_this` by aligning it more with how Svelte 4 bindings work, namely reacting to each block context variables updates properly and introducing a mechanism to get the previous binding destination using said variables.
* fix: improve $inspect handling of derived objects
* Update packages/svelte/src/internal/client/runtime.js
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
---------
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
* fix: remove memory leak
* changeset
* mutable_source is only used in non-runes mode, no need to check here
* simplify
* oops
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* fix: remove readonly validation
The readonly dev time validation results in false-negative object equality checks: The original object is proxified and thus comparisons could be between the unproxified and proxified version, which will always return false. Fixes#10372
There's also the problem that an object could be passed into a component but then passed upwards into shared state. At that point, it should no longer be readonly, but it's not possible to statically analyze these points. Fixes#10372
Lastly, the each block logic mutates an internal array and that also throws errors with the readonly logic. Fixes#10037
* reinstate tests
* track ownership of state and mutations
* working?
* remove old changeset
* tidy
* error
* simplify
* fix
* fix
* fix
* tidy
* make it a warning
* rename test
* remove unused test
* update tests
* slap ts-expect-error everywhere, because its too finicky otherwise
* oops
* oh no the hall monitor is here
* only call add_owner in dev
* only owners can transfer ownership
* simplify
* fixes
* tidy up
* fix type error
* while we're at it
* rename file
* rename functions
* add some comments
* move ownership checking logic
* ugh eslint
* more detailed message
* only add filename in dev
* comment
* update tests
* move more code
* undo change to sourcemap tests
* allow proxy to have multiple owners
* use SignalDebug
* i was doing this all wrong
* tidy up
* implement inheritance
* fix
* tidy up
* update filename stuff
* changeset
---------
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
Co-authored-by: Rich Harris <rich.harris@vercel.com>
Adjusts the escaping mechanism done for server compilation. For template literals it's now only applied when explicitly told, which is the case for generated literals from the html template. Fixes a bug where a template literal string inside the `@html` tag was wrongfully escaped (https://github.com/sveltejs/svelte/issues/10359#issuecomment-1949678228)
In Svelte 4, scripts inside `@html` were not executed when it was created client-side. This is because `innerHTML = ..` which was used under the hood does not execute scripts due to security reasons. This adjusts the code so the same is true for Svelte 5.
Ensure update methods of actions and reactive statements work with fine-grained `$state` by deep-reading them. This is necessary because mutations to `$state` objects don't notify listeners to only the object as a whole, it only notifies the listeners of the property that changed.
fixes#10460
fixes https://github.com/simeydotme/svelte-range-slider-pips/issues/130
* failing test for :is(...)
* simplify
* trim
* factor out truncate function
* pass stylesheet around
* recurse into :is and :where
* fix types
* fix types
* almost there
* gah so close
* tada
* changeset
* simplify
* feat: nested CSS support (#10491)
* parse nested CSS
* tests
* track parent rules
* some progress
* switch it up
* pruning
* works
* changeset
* lint
* error early on invalid nesting selector
* tidy
* note to self
* fix some specificity stuff
* failing test
* note to self
* fix: correctly scope CSS selectors with descendant combinators (#10502)
* fix traversal, but break some other stuff
* man this is fucken hard
* fixes
* getting closer
* be conservative for now
* tidy
* invert
* invert
* simplify
* switch
* for now
* progress
* i think it works?
* fix
* tidy up
* revert some stuff
* remove some junk
* handle weird cases
* update
* tweak
* shrink
* changeset
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* Update playgrounds/sandbox/run.js
* changeset
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
* feat: add hydrate method, make hydration treeshakeable
Introduces a new `hydrate` method which does hydration. Refactors code so that hydration-related code is treeshaken out when not using that method.
closes#9533
part of #9827
* get docs building
* ugh
* one more
* Update packages/svelte/scripts/check-treeshakeability.js
Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
* warn
* Update sites/svelte-5-preview/src/routes/docs/content/01-api/05-functions.md
---------
Co-authored-by: Rich Harris <richard.a.harris@gmail.com>
Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <rich.harris@vercel.com>
Add source map merging for preprocessors and get tests passing.
- fixed some issues around the `sources` array where they weren't calculated relative to the input correctly
- adjusted some CSS tests because due to our own CSS parser the AST is less granular, so there are less mappings now. Don't think this is a problem, but worth thinking about
- removed enableSourcemap but only log a warning, the reason this was introduced was to mitigate a bug in Vite which occured when having the source map inlined into the SSR'd output. Since SSR doesn't contain inlined CSS anymore (and if it did, we would omit the source map) the reason for which it was introduced no longer exists
- files without js mapping in it have no source mappings yet (originally added in Svelte 4 for #6092)
* rename $derived.call to $derived.by
* more replacements, plus a compiler error
* regenerate types
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
Relax the runtime error to only throw when you're passing a binding with an undefined value. This makes it possible to provide components in a way that can be used more flexibly while keeping the error to guard against the case that we want to avoid: a default value propagation up.
* fix: handle sole empty expression tags
When there's only a single expression tag and its value evaluates to the empty string, special handling is needed to create and insert a text node
fixes#10426
* fix
* need this, too
* Update packages/svelte/src/internal/client/operations.js
---------
Co-authored-by: Rich Harris <richard.a.harris@gmail.com>
Closes#9420.
This PR creates an $effect.pre (before beforeUpdate and an $effect (for afterUpdate) and, inside those, listen for all locally declared signals plus reactive props. This does mean that we need to link the locally declared signals to the component context (the reverse of the current behaviour, wherein we link the component context to locally declared signals).
fixes#10296
Also make sure to use the server export conditions when resolving Svelte imports from inside the server compiler output
---------
Co-authored-by: Rich Harris <rich.harris@vercel.com>
Co-authored-by: Simon Holthausen <simon.holthausen@vercel.com>
- Previously, any each block parents where used as keys for the sub group. That's wrong, it should only be using each blocks whos declarations contribute to the `bind:group` expression. Fixes#10345
- Only the left-most identifier of the expression was used to determine the binding group. This is wrong, all identifiers within the expression need to be taken into account. Fixes#9947
Paolo Ricciuti
Caique Torres
Simon H
Rich Harris
github-actions[bot]
Craig Jennings
Yuichiro Yamashita
Vaibhav Rai
Jeremiasz Major
Tan Li Hau
Dominic Gannaway
Daniel
Frédéric Crozatier
miunau
Rich Harris
Mostafa Kheibary
Ahmad
xxkl1
Daniel Imfeld
Simon Holthausen