Cancel the full thumbnail upload/render stream on page or workspace
teardown and on matching thumbnail clearing. Deferred library thumbnails
can otherwise query a main missing from the next WASM scene and abort
the renderer (F112).
Complete the upload observable after its callback so successful requests
also release their subscriptions.
Add async regressions for pending current and library frames, late upload
callbacks, clearing, and successful rendering. They fail before the fix.
Document the thumbnail lifecycle in the frontend memory.
Frontend CI passes lint, 1499 tests / 7272 assertions, and formatting.
Live WASM update-main followed by navigation no longer crashes; an
uninterrupted library request refreshes its PNG thumbnail too.
AI-assisted-by: GPT-6
Keep a namespaced file destination on each history change. Replay changes
in order in their target files, including stacked entries and transactions.
Keep payload fields named file-id intact, such as set-tokens-source (F111).
Snapshot the current file and every library in both composable runners.
Name the file in failure signatures; normalize the server synced-at status
stamp alongside modified-at. Update golden cases H and AP for this exclusion.
The library update regression checks both files through undo and redo and
failed against the old code.
Common JVM: 1705 tests; JS: 1633; frontend: 1495; backend: 1610; exporter: 14.
All pass, with lint and formatting checks for both changed modules.
Benchmark vs Phase 1e (lower quantile, ms):
world baseline now ratio
main-10 1.790 0.993 0.55
main-100 15.415 9.038 0.59
main-1000 153.086 90.145 0.59
three-levels 35.215 18.989 0.54
chain 68.931 35.045 0.51
library-500 83.083 85.830 1.03
many-shapes 4.837 3.344 0.69
variant-100 32.052 33.586 1.05
All worlds pass the 10% regression threshold.
AI-assisted-by: GPT-6
Emit changes-applied with the commit's prior file data. Sync each commit
before a later edit can replace its state or separate its undo group.
Allow later commits through while completion barriers finish (F110).
Process layout batches after their buffer closes, so reentrant component
sync cannot lose a pending layout task. Cover three user edits without an
intermediate wait, then every undo and redo. Token and tiny-layout cases
also pass. Re-record golden cases X and AJ for the settled timing.
Frontend: 1495 tests, 7255 assertions pass; lint and format checks pass.
The three-edit regression failed before the fix.
Benchmark vs Phase 1e (lower quantile, ms):
world baseline now ratio
main-10 1.790 0.993 0.55
main-100 15.415 9.038 0.59
main-1000 153.086 90.145 0.59
three-levels 35.215 18.989 0.54
chain 68.931 35.045 0.51
library-500 83.083 85.830 1.03
many-shapes 4.837 3.344 0.69
variant-100 32.052 33.586 1.05
All worlds pass the 10% regression threshold.
AI-assisted-by: GPT-6
Decision (c) of the component contract, after a fix that made the sync of a
library file reach that file at all (Phase 3c).
- Commit the sync of a library file to that file.
An update of the main of a library component, made from a copy in the
file, syncs the copies of the component in the file and in the library.
The changes of a sync carry no file id, so the commit took the current
file's: the library's copies, among them those its other mains nest, were
written to pages the current file does not have, and never changed (F19).
A sync now commits to the file it syncs.
workspace-remote-cascade-test: an update of the main of a library
component reaches the copy of it that another library main nests.
- Sync the components a sync changes in the same commit.
Decision (c) of the component contract: a sync reaches the copies of the
components it changes, until none is left, in one commit. A sync of a file
changed the mains of the components that nest the synced copies, and the
component watcher synced those in a commit of its own, one level per pass,
so the file stayed partly synced in between; the watcher only watches the
current file, so a sync of a library file stopped at its first level
(F19).
common: generate-sync-file-cascade syncs a file as
generate-sync-file-changes does, then the copies of each component of the
file that the sync changed, one per pass, applying each pass to the file
data, until none is left (generate-sync-components-cascade). The
composable runner, the benchmark and the golden master use it.
frontend: sync-file commits the whole cascade at once. When it syncs a
library file, it then syncs the copies in the current file of the library
components the cascade changed.
workspace-remote-cascade-test: an update of the main of a library
component, made from a copy in the file, reaches the file's copy of a
library component whose main nests it. The common golden master records
fewer changes undone in case W: a cascade is one change set.
Benchmark (lower quantile, ms; baseline / now / ratio):
deepest-edit-through-three-levels 35.2 / 17.0 / 0.48
edit-through-a-chain-of-dependents 68.9 / 33.2 / 0.48
library-update-into-copies 83.1 / 43.6 / 0.52
main-edit-among-many-shapes 4.8 / 3.5 / 0.73
main-edit-into-10-copies 1.8 / 0.9 / 0.49
main-edit-into-100-copies 15.4 / 8.1 / 0.52
main-edit-into-1000-copies 153.1 / 83.9 / 0.55
variant-switch-into-copies 32.1 / 19.6 / 0.61
AI-assisted-by: claude-opus-5-5
Decision (f) of the component contract, with a fix it needed and a check of
decision (g) (Phase 3c).
- Return empty modifiers for a scale constraint axis.
The constraint of each axis of a child gives the modifiers that the
parent's resize adds to the child; :scale has no method of its own and
took the default, which returned an empty vector. Adding a vector to
modifiers reads its :last-order as nil: JavaScript adds nil as 0, the JVM
throws, so a child scaled on one axis and constrained on the other (scale
and top, say) could not be resized on the JVM. The default now returns
empty modifiers.
geom-shapes-constraints-test checks the default is empty and adds to
modifiers.
- Keep the children of a transformed copy shape derived.
Decision (f) of the component contract: transforming a copy shape touches
that shape only; its children change as a consequence and stay untouched,
and sync gives them their expected geometry, their main's placed against
their parent and resized through their constraints.
Touch. The placement step decided per shape whether a transform touched it
by comparing it with its copy root, so the children of a resized root that
stretched (scale, left and right) were touched (F10), a derived width or
height always touched (F22), and a resize that scales the content touched
the strokes, radii and the like of every child (F39). The editor now tells
the placement step which shapes the user transformed (the keys of the
modifiers given, or those the stored modifiers were set for); any other
shape of the transform changes only as a consequence of theirs (a child
through its constraints, a board that hugs them) and touches nothing.
Sync. Sync placed every main shape by the offset between the roots only,
so the children of a resized copy root that a resize leaves untouched
(center) went back to their main's place at the next sync of the main
(F10). Now, before it walks the children of a copy shape that keeps its
own geometry, sync resizes, turns and moves the main shape's subtree to
match it, its children following through their constraints and layouts,
and walks those: each child gets that geometry unless it has its own.
comp-expected-geometry-test: a copy root resized as the editor does places
a child constrained left, right, center, scale and left-right, leaves it
untouched and keeps it there through a sync; a resize that scales the
content scales the child's stroke and touches only the root. The frontend
golden master records, in cases AI and the touched child under a main
rotation, the children the resize or rotation of a root no longer touches.
- Check a copy child resize leaves its hug board untouched.
Decision (g) of the component contract: the size of a board that hugs its
content is the layout's, never an override. The layout reflow writes it as
derived, and a transform touches only the shapes the user transformed
(decision (f)), so a resize of a copy child that grows the board it sits
in touches the child only.
workspace-layout-owned-sizes-test pins it.
Benchmark (lower quantile, ms; baseline / now / ratio); the machine
was loaded, and the code before this change, run in between, gave the
same spread (0.44 to 0.59):
deepest-edit-through-three-levels 35.2 / 16.7 / 0.47
edit-through-a-chain-of-dependents 68.9 / 33.4 / 0.48
library-update-into-copies 83.1 / 40.3 / 0.49
main-edit-among-many-shapes 4.8 / 2.4 / 0.49
main-edit-into-10-copies 1.8 / 1.0 / 0.54
main-edit-into-100-copies 15.4 / 8.8 / 0.57
main-edit-into-1000-copies 153.1 / 92.7 / 0.61
variant-switch-into-copies 32.1 / 19.6 / 0.61
AI-assisted-by: claude-opus-5-5
The name a text takes from its content and the size it is measured to are
derived: they never touch a copy (decisions (d) and (g), Phase 3c).
- Let a copy text's automatic name follow its own letters.
The editors name a text after its letters on every edit. In a copy that
rename went through the same write as the content, so it touched
:name-group (F27), against decision (d) of the component contract: only an
explicit rename is an override.
- update-shapes takes :ignore-touched as a set of attributes too, and the
editors' commits (v1 and v3, finalize and keystrokes) and the replace of
text pass #{:name}: the name follows the content, so its write is
derived. Token unapply still runs for such an update.
- Sync gives an untouched copy text the name generated from the letters it
shows after the sync when its main's name is the one generated from the
main's letters (decision (d), refined), and the main's name otherwise.
So a copy that retyped its text keeps that name, and one that did not
follows the main's.
The pure runner's text edit makes the same commit. Case Y no longer pins
F27 and checks the copy's name follows its letters; both golden masters
record the copy without :name-group.
text-sync-test: an untouched copy takes the main's letters and name, a
copy that typed its letters keeps the name they give, and a main named by
hand passes its name on.
- Name a text with several paragraphs without line breaks.
The editors name a text after its letters, with the line breaks of its
paragraphs included, so a text "Hello world" + Enter + "Ends Sunday" was
named "Hello world\nEnds Sunday"; the layers panel showed a space there
(F82). generate-shape-name now turns each line break, with the spaces
around it, into one space.
text-test: the name of a two-paragraph text joins them with a space.
- Keep the size a WASM text grows to from touching the copy.
A text that grows with its content (auto width, auto height) gets its size
from the WASM renderer after a font loads, a style changes, a paste or an
edit ends. That size is derived from the content, yet the resize applied
it as a user change, so a copy's text took :geometry-group each time and
kept that size against its main (F1). The resize and the debounced resize,
the size an edit ends with and its geometry undo now pass ignore-touched.
These paths measure the text in WASM, so no headless test reaches them.
Checked in the browser with the WASM renderer: an edit of a copy's text
touches :content-group only, and a resize after a font size change of 24
grows the text with no :geometry-group.
- Keep the finalize resize of a text out of the undo stack.
When an edit of an existing text ends, the WASM renderer gives the size it
grows to, and the finalize commit stores that size along with its undo.
The resize that shows it on the canvas recorded an undo of its own with no
undo group: a release build kept a second undo entry for one edit, and a
dev build failed the undo entry check (F36). It now records none; a new
text still joins the undo group of its creation.
Checked in the browser with the WASM renderer: an edit of an existing text
adds one undo entry.
- Keep the size a text edit ends with from touching the copy.
When an edit of a text ends, the editor stores its content together with
the size measured for it (the SVG renderer's text modifier, or the WASM
measure in the v1 editor). Only the name was written as derived, so an
edit of a copy's text touched :geometry-group besides :content-group, and
the text kept its size against its main (F1, in the SVG renderer). These
writes now take the size, the position data and the name as derived; the
content still touches.
workspace-text-modifier-test: an edit that ends with a measured width
touches the copy text's content only. Checked in the browser with the SVG
renderer: an edit of a copy's text touches :content-group and
:text-content-text only.
- Measure again on WASM the texts a sync writes.
The size of a text that grows with its content is derived (decision (g) of
the component contract): sync writes the main's, and the renderer measures
the text again. The SVG renderer does so when it draws the text; the WASM
renderer did not, so a copy text with content of its own kept its main's
width and height after a sync (F35). Now, with the WASM renderer, a sync
of the current file measures again the auto-grow texts of the current page
it writes, without a component sync of its own, in the undo group of the
sync; a sync with no undo group gets one for both.
Texts on other pages keep the size sync wrote until they are measured
again.
workspace-sync-wasm-text-test: a sync that writes a copy text with content
of its own measures it again, in the undo group of the sync. Checked in
the browser with the WASM renderer: after the main text grows or shrinks,
two copy texts with content of their own get their own widths and no
:geometry-group.
Benchmark (lower quantile, ms; baseline / now / ratio):
deepest-edit-through-three-levels 35.2 / 18.1 / 0.51
edit-through-a-chain-of-dependents 68.9 / 34.0 / 0.49
library-update-into-copies 83.1 / 40.2 / 0.48
main-edit-among-many-shapes 4.8 / 2.5 / 0.51
main-edit-into-10-copies 1.8 / 0.8 / 0.44
main-edit-into-100-copies 15.4 / 8.5 / 0.55
main-edit-into-1000-copies 153.1 / 83.5 / 0.55
variant-switch-into-copies 32.1 / 18.1 / 0.56
AI-assisted-by: claude-opus-5-5
Fixes in when the workspace syncs copies and propagates tokens, and in what
a layout wait covers (Phase 3c).
- Sync the copies when a shape is dragged inside a main.
The component watcher dropped every commit flagged :translation? (the WASM
drag commits only moves) so dragging whole mains around the canvas stays
fast. A drag of a shape inside a main is flagged the same, so its copies
never followed until some other edit of that main made them jump (F70).
The watcher now drops a translation commit only when it moves shapes
outside mains or whole mains: every shape it moves inside a main moves
with the main's root (decision (q)). Otherwise the commit goes through
component sync like any edit.
workspace-sync-barrier-test: a translation of a whole main is dropped; one
of a main's child syncs the copy. fill-commit in the interpreter takes
several labels.
- Propagate the file's tokens after any library update.
Updating a library from the notification synced the file and then
propagated its tokens, so the copies take the token values of the file's
token status, not the library's. The update button of the libraries
dialog's Updates tab only synced the file (F83).
dwl/update-libraries syncs the file from the given libraries and then
propagates its tokens; the notification and the Updates tab both use it.
workspace-library-updates-test: updating a library syncs it and then
propagates the tokens.
- Sync copies whether or not component thumbnails are on.
The component watcher's whole pipeline, the sync of the copies included,
ran only when the :component-thumbnails flag was on or the WASM renderer
was active. The flag is on by default, but an install that turned it off
never synced copies on the SVG renderer: editing a main left its copies as
they were (F8). The flag now gates only the thumbnail streams.
workspace-sync-barrier-test: with the flag off, a main edit reaches the
copy.
- Hold a token's shapes pending until the token is applied.
A plugin's waitForLayoutUpdate waits until no shape has layout work
pending. Applying a token resolves the tokens first, asynchronously, and
opened no pending work meanwhile, so the wait found nothing pending and
returned before the token reached the shapes or their copies (F54). The
shapes are now pending, as a :token task, from the start of the resolution
until the update is emitted, and the layout, text and sync work it starts
takes over from there.
The composable interpreter no longer waits on the token itself: case AY
applies a token and checks its copies after the same wait a plugin uses,
and fails without this change.
AI-assisted-by: claude-opus-5-5
Two derived commits that left the copies' sync in an undo entry of its own
(Phase 3c).
- Move frame guides in the undo group of their transform.
A transform of a board sends the move of its guides before the shapes'
update. The guide move committed every guide of the board, moved or not,
each in a commit with an undo group of its own, so when the board was a
main with a guide, its transform and the copies' sync landed in different
undo entries: one undo reverted only the copies (F6).
Both transform paths now give the shapes' update an undo group up front
and pass it to the guide move, which commits only the guides that move,
all in one commit in that group.
workspace-frame-guides-undo-test: one undo reverts a rotation of a main
with a guide, its copy included, and a move of a main together with its
guide.
- Reflow layouts in the undo group of the step that asked for it.
A layout reflow runs a moment after the commit that needs it: the
:layout/update events are buffered for 100 ms and reflowed per page. The
reflow lost the undo group of the step (F43):
- the buffer passed no undo group to the reflow;
- update-shapes asked for a reflow without the group of its commit;
- the reflow's transform opened an undo transaction, so it made an entry
of its own instead of stacking on the latest one, and the sync that a
reflow of a main starts took that entry's fresh group.
So one undo after a sync that reflows (a flex main's margin edit, a nested
sync with minimal frames, a swap) reverted only the reflow, or only part
of the step.
update-shapes now gives a commit that needs a reflow an undo group up
front and passes it on; the buffer reflows each page in the group an
update of that buffer names (those that name none come from the same step,
like update-layout-child's); the reflow stacks on the latest entry with no
transaction of its own.
Cases AJ and AN no longer pin F43: one undo after a nested sync reverts
the remote change and the float noise of the reflow, and one undo after a
flex main's margin edit reverts the margin, the sync and the reflow. The
frontend golden master records that.
AI-assisted-by: claude-opus-5-5
Fixes in the variant events of the workspace and in the component calls of
the plugin API (Phase 3c).
- Rename a variant in one commit.
Renaming a variant renamed its board, then each of its components and
mains, as separate commits inside an undo transaction. The transaction
kept the undo group of its first commit, the board's, while the sync that
the renamed mains started took the group of their commits and made an
entry of its own: one undo reverted only the copies' names (F6, case AV).
clv/generate-rename-variant builds the board, component and main renames
in common, and the event commits them once, so the rename and its sync are
one undo entry, like transform into a variant.
Case AV no longer marks its round trip as a known failure; the frontend
golden master records it without failures.
- Keep overrides in a plugin swap only within a variant set.
The plugin's swapComponent swapped with keep-touched for any component,
while the UI keeps a copy's overrides only on a variant switch: the same
swap kept different overrides by route (F33). It now keeps them only when
the copy's component and the target are variants of one set (decision
(p)), as swap-keeps-touched? says.
shape-bugfixes-test: two variants of one set keep the overrides; of
different sets, or plain components, do not.
- Look up a plugin variant head's component in its own file.
The plugin's isVariantHead looked the shape's component up in the file of
the shape, so it was false for copies of a library's variants; and it
looked it up before checking the shape is a head, so on a shape with no
component it failed the lookup's assert (F91). It now checks the head
first and looks in the shape's :component-file.
shape-bugfixes-test: a copy of a library variant is a variant head, and a
plain rect is not.
- Add a variant with the shortcut when a variant set is selected.
The context menu of a variant container offers "Add variant" and shows the
create-component shortcut for it, but the shortcut's event did not know
containers: with one selected it fell through to creating a component,
which the container's own rules refuse (F95). With a single variant
container selected, the shortcut now adds a variant to it, as the menu
entry does.
workspace-variants-test: the shortcut on a variant container gives the set
a third member.
- Refuse plugin variant edits whose variant is on another page.
The variant events the plugin API calls (add a variant, add, remove or
rename a property, set a property value) build their changes on the
objects of the page the workspace shows. For a variant whose mains sit on
another page, they updated the components but not their mains, so the
mains kept their old :variant-name and the file failed validation (F94).
The plugin now refuses these edits, as it refuses edits of shapes on
another page, unless the variant is in the current file and on the current
page.
library-test: each edit goes through with the variant on the current page
and is refused, with nothing emitted, with it elsewhere.
- Check every component of an assets group before combining it.
The assets panel offers to combine a group of components as variants when
they can be: more than one, none a variant, all with their main on one
page. It counted every component of the group, subgroups included, but
checked the other two rules only on the components directly in it, while
the combine takes them all; with none directly in it, those rules always
held (F98). can-combine-group? now checks every component of the group.
assets-components-test: a group whose only components are in a subgroup
can be combined; a variant or a main on another page in a subgroup blocks
it; one component cannot.
AI-assisted-by: claude-opus-5-5
- The UI offers swap and reset only on copy heads, and detach only on
copies that are not inside another copy. The plugin's swapComponent
and resetOverrides accepted any shape inside a copy (F92), and its
detach any shape at all (F93), and sent the same events, whose
changes expect a head. swapComponent and resetOverrides report "The
shape is not a component copy head" for anything else; detach
reports a shape that is not a copy, or a copy inside another copy,
by ctn/detachable?, which the detach event checks too. swapComponent
looked up the target file before checking its argument, so a value
that is not a component failed an assert instead of reporting
"Component not valid".
- The plugin's switchVariant did not check whether the switch would
nest a component in itself. The switch event checks that and calls
the :on-error of its params' metadata; the plugin passed none, so
the default, rx/throw, built an error observable that nothing used,
and the call did nothing and reported nothing (F53). The plugin
passes an :on-error that reports "The switch would create a
component nesting loop", and the event's default does nothing.
- Choosing a property value in the design tab with several members
selected sent, for each member, an event that set the value and
another that cleared its error, each its own commit and undo entry.
The plugin's setVariantProperty set the value and left the error as
it was (F94). dwv/update-property-value takes the components, and
sets the value and clears the error of each in one commit; the
design tab passes the selected members, the plugin its component,
and the composable interpreter the one member of its edit.
shape-bugfixes-test checks each plugin case on the proxy, and
workspace-variants-test sets a value on two members with errors.
AI-assisted-by: claude-opus-5-5
Several actions built their changes in frontend events, step by step,
each step its own commit, so the pure runner could not run them and
one undo could revert half an action.
- Renaming a variant member in the layers panel sent three events,
each in its own undo transaction: update the properties from the
typed formula, remove the properties no member gives a value, and
set or clear the member's error. One undo reverted only the last.
clvp/generate-rename-member builds the three steps in one changes
(with generate-update-properties and
generate-remove-empty-properties), and dwv/rename-member commits it
once with the same analytics events.
update-properties-names-and-values goes; remove-empty-properties and
update-error stay for the design tab.
- Transforming a component into a variant sent about fourteen events,
and combining components as variants that transform and eight more.
clv/generate-transform-in-variant and
clv/generate-combine-as-variants build those steps as one changes,
from the generators common already had, and the events commit them
once. The flex layout initializer moves to common as
cls/layout-initializer. Combine still sizes the board with two
frontend dimension updates inside its undo transaction: the
frontend applies them through modifiers, which common has no step
for.
- update-component split the changes of an update main (the copy's,
marked :local-change?, from the main's, which may belong to a
library) with two transducers; cll/split-local-changes does it. Its
undo-group argument, never passed, goes, and the main's library is
the copy head's :component-file, which it looked up through the
file data only to read its id back.
- dwv/valid-variant-switch? read the libraries from the state and
repeated the lookup of clv/find-switch-target. clv/valid-switch?
takes the libraries and asks find-switch-target; the plugin's
switchVariant and the frontend test call it.
Behavior changes:
- One undo after a rename of a member, a transform or a combine
reverts all of it. After a transform, it reverted only the sync of
the copies' names (F6 in case AT); its pins go.
- The transform no longer sends the "create-board" analytics event of
the board it creates around the main.
In the composable suites, renaming a member in the layers panel,
removing the empty properties and setting a member's error are
variant property edits both runners run, and case AW joins the JVM
set. Its rename node drives the real layers-panel rename, which
removes the empty properties too, so the step that removed them goes.
The golden masters are recorded again for AW (the JVM one gains the
case, the frontend one loses that step) and for AT (its final states
no longer list the F6 undo failures).
AI-assisted-by: claude-opus-5-5
remove-empty-properties, run after a rename in the layers panel, found
the position of each property by grouping all the properties of the
variant's components by name and numbering the groups. A map of more
than eight keys does not keep their order, so with nine or more
properties the numbers did not match the positions and the wrong
property was removed: with ten properties and the sixth empty in every
component, it removed the first (F86).
The positions now come from the properties themselves: a property is
removed when its value is empty in every component. The work moves to
common as clvp/generate-remove-empty-properties, and the event only
commits what it builds.
workspace-variants-test checks the case with ten properties.
AI-assisted-by: claude-opus-5-5
The clear-thumbnail-batch tests mock the RPC and count every call
recorded while the mock is installed. A thumbnail deletion batch left
over by an earlier test of the suite (a component sync clears the
thumbnails of the frames it changes) can call it in that window, so
the tests failed in the full frontend run from time to time and passed
alone.
- The batch tests count the calls about their own object ids.
- The test that clears an empty queue expected no call at all; a call
made by the event under test could only carry no object ids, so it
checks that none of the recorded calls is such a call.
AI-assisted-by: claude-opus-5-5
Sync and the variant switch can take a copy text's family from one side
and its weight from the other. The variant id goes with the weight, so
it can name a variant of the other family: the WASM renderer then picks
the closest variant, but the SVG renderer falls back to the regular one.
The frontend now gives each text content that a sync or a component swap
commits variants of its own families: a run whose variant is not one of
its family's takes the family's closest to its weight and style
(cttx/update-font-variants, fonts/find-closest-variant). The pure runner
has no fonts, so case BA leaves variant ids out of its comparison.
The merge also leaves alone a run that already shows the placed style,
so a main edit no longer rewrites the variant ids of a copy whose
family is its own; with a new family, the run still takes the placed
variant id.
Benchmark (lower quantile, ms; baseline / now / ratio):
deepest-edit-through-three-levels 35.2 / 35.7 / 1.01
edit-through-a-chain-of-dependents 68.9 / 65.2 / 0.95
library-update-into-copies 83.1 / 75.1 / 0.90
main-edit-into-10-copies 1.8 / 1.5 / 0.86
main-edit-into-100-copies 15.4 / 15.1 / 0.98
main-edit-into-1000-copies 153.1 / 151.4 / 0.99
variant-switch-into-copies 32.1 / 30.4 / 0.95
AI-assisted-by: claude-opus-5-5
Both binfile importers stored a file without the components/v2 feature
as a v2 file, though its component data was never migrated. Such a file
is now rejected with :file-in-components-v1, the code that opening a v1
file already raises, when it has components or shapes linked to them
(:component-id, :shape-ref or :main-instance). A file without component
data takes the feature and imports as before.
Since every imported file now has the feature before the pending
migrations are registered, the no-op components/v2 migration goes.
The import dialog explains the rejection: the entry says that the file
uses an old type of components that Penpot no longer supports, with its
own text, as errors.deprecated talks about opening a file.
AI-assisted-by: claude-opus-5-5
A review found tests that would pass with what they test broken.
- Runners: both judge one variant at a time, fail a variant with no
assertion and a known-failure mark that applies nowhere, and give
the asserter the state the operations produced. The pure runner
validates the final file and rejects operations of unknown kind.
The frontend interpreter fails a step when the store reports an
event error (Potok only logs it), reports asserter errors against
their case, and stops the watchers that install! starts.
- Nodes drive the real code: a main root rename goes through
end-rename-shape, and the clipboard's advance of nested copies is
one common function, cll/advance-copied-objects, that copy-selected
and both test paths call.
- Cases N, AH, AI, AK, AD, AE, K, C, H, S, AY, I, Y, AX and F check
what they claim; O, R and T get real sizes.
- Stricter common unit tests found F60 (a repair detaches healthy
copies), F61 (a switch drops a fixed-sizing size), F62 (the ref
chain never stops at its target) and F63 (latent); all pinned.
- Frontend library tests: the RPC double records on subscribe, the
dropped commits check that nothing syncs, and the link and share
tests check the events sent.
- Plugin suite: OpSwap and OpDetach throw when the API call did
nothing, and case AB checks that the reset head is an instance of
the original component.
- The composable suite memory records the new rules.
The composable interpreter treated a step as done after a 60 ms gap
with no commits, and treated its 2 s timeout as done too. Slow sync or
reflow work could finish after the step, so a case could assert on a
half-synced file.
Each event step now waits, as the plugin waitForLayoutUpdate does,
until every pending work tracked by the reflow registry has drained:
the watcher's sync wait, the sync it starts, and layout and text
reflow. A step that does not drain within 2 s fails the test and names
the work still pending, and the next variant starts with nothing
pending.
The tests cover an edit with a deliberately slow sync, undo and redo
steps that must wait for the watcher's touch-component, translation
and derived commits, a sync that never finishes, and a restart of the
edit watchers, which must keep the undo entries of an edit the same.
Also in this commit:
- Keep page-less changes from adding a nil page. Applying a commit locally
rebuilds the object indices of every page its changes name. Changes that touch
no page, such as :mod-component, name none, and their nil page id was rebuilt
too: the file data in the store gained a page under a nil key holding
{:objects nil}. Every component touch after an edit to a main added it, and
undo never removed it.
- Check undo and redo of every step in the composable cases. The composable
cases tested undo only where a case asked for it, so a change generator with a
wrong inverse, or a step that splits into several undo entries, went
unnoticed.
- Add operations and cases to the composable component suites. The composable
runners could edit, add, remove and move children, swap and switch, but not
run the other component operations.
- Test component validation, repair and library asset sync.
- Add a component sync benchmark with a recorded baseline. The coming refactors
and fixes change sync code, and a slowdown must be seen and explained in the
commit that causes it.
- Tidy the composable suite and make its cases stricter.
- Test the component sync paths no test reached. A coverage run found sync paths
that no test ran.
- Test variant events, token sync and switch overrides. The variant events had
almost no frontend tests, and several switch, paste, token and restore paths
never ran.
- Test the file and container helpers and the library events. The component
helpers of types/file and types/container and the library events of the
workspace had the lowest coverage left.
The component watcher opened its :sync-file wait only for commits saved
to the undo stack. Undo and redo commit without it, so a wait such as
the plugin waitForLayoutUpdate returned before the watcher had handled
them. Translation and derived commits were filtered out before the wait
opened.
Open the wait for every local commit of a file and close it once the
watcher has decided what to do: sync, only touch the component, or
drop the commit. Dropped commits close it right away.
Add wrf/pending, a snapshot of the pending work, for tests and
diagnostics.
* 🐛 Require thread access when marking threads as read
mark-all-threads-as-read took any list of thread ids and wrote the read
status of the caller on each of them, without checking that the caller
can access the file that owns the thread. It was the only comment
mutation that did not check permissions, and it took the ids as an
unbounded array, so a single request could keep a database connection
and a transaction open for a long time.
The rpc now resolves the threads in one query and checks the comment
permissions of every distinct file, answering not-found when a thread
does not exist or is not accessible. Both cases answer the same error
on purpose, so the response cannot be used to tell existing threads
apart.
The status rows are now written with a single insert, keeping the
modified-at date one second in the future that makes the thread count
as read. Repeated ids are deduped, the command returns nil again like
the sibling commands with no result, and the schema limits the number
of threads per call to 500 while the dashboard sends the ids in batches
of that size.
Tests cover access rejection, unknown threads, mixed batches that write
nothing, repeated ids, empty lists, marking twice, and the frontend
batch split.
Relates to GHSA-j4vc-cmf6-jhr7
AI-assisted-by: minimax-m3.1-flash
* 🐛 Refresh the unread list when a batch fails
The dashboard sends the thread ids in batches of 500 and each batch
commits on the server. A failure in a later batch left the unread list
stale: the re-read only ran when every batch succeeded, so the threads
already marked as read stayed visible as unread.
The re-read now runs in both paths. On failure the event is emitted
before the error is propagated, so the error toast still shows and the
success toast stays limited to a full success.
The backend also answers an empty list without opening a transaction,
since there is nothing to check and nothing to write.
Relates to GHSA-j4vc-cmf6-jhr7
AI-assisted-by: deepseek-v4.1-flash
* 🐛 Restrict the team invitation token and list to admins
The token of a team invitation is a bearer credential: whoever holds it
accepts the invitation with the role of the invitation row. The RPC that
mints it only required edition permissions, while the rest of the
invitation commands require admin, so a team editor could get the token of
any pending invitation, including the ones that grant the owner role.
The token now requires admin, and the token of an owner invitation is
restricted to the team owner: an admin cannot create nor promote an owner
invitation, so it cannot keep its token either. It reuses the error code
already used for that rule.
Minting the token now leaves an audit event named as an action, carrying
the granted role and the id of the invitation, so the extraction does not
go unnoticed.
The list of pending invitations is admin only as well: it carries the
email of the invited person and the role offered, and the dashboard was
painting those emails to any member. The frontend catches the permission
error and shows the permission notice instead, and the copy action is not
offered for an owner invitation, which the backend rejects for an admin.
The five invitation commands now answer with the same permission error,
so they speak one language.
Relates to GHSA-4mmc-j8rh-74pf
AI-assisted-by: minimax-m3.1-flash
* 🐛 Clear the cached invitations and gate the list on team admin
The pending invitations are a team admin only list. The dashboard still
offered the entry to every member and keyed the section on the invite
policy of the organization, so a team admin blocked by that policy saw
the no-permission notice although the backend lets them list.
- `fetch-invitations` answers a denied refetch with an
`invitations-fetched` event carrying an empty list, so a demoted admin
does not keep the cached emails and the header no longer paints the
invite button from a stale list.
- The header tabs and the team options dropdown hide the invitations
entry without `:is-admin`, which already includes the team owner.
- `invitations-section-content` keys the section on `:is-admin`, and the
invite policy only gates the invite actions: resending keeps the
`can-invite?` check, deleting and row selection only need team admin.
AI-assisted-by: deepseek-v4.1-flash
When components whose name matches the variant container are moved into
it, removing the container name leaves an empty value. Combining several
components named the same produced "Value 1" for the first variant and
empty values for the rest, so variants shared the same combination.
Assign the first unused "Value N" to the last property when all computed
values are blank, tracking values given within the same batch.
Combine moves the remaining mains into the container one at a time, in
the combine order, so the generated values follow that order instead of
the layer order.
Closes#12878
* ⚡ Keep the translation flag on undo and redo of a move
Previously undos and redos always forced a component sync + thumbnail
rerender for every component touched even if the underlying commits were just
translations. This propagates the translation flag to undos and redos and
propagates it to commit-changes. An entry merged from several undos/redos is
a translation only if all of its undos/redos are translations.
This reduces a UI freeze when undoing a move of a board with 453 components
from 2.42s -> 0.22s.
AI-assisted-by: claude-sonnet-5-5
* :translation? -> :translation
---------
Co-authored-by: Andrey Antukh <niwi@niwi.nz>
* 🐛 Return the full profile from the LDAP invitation branch
The invitation branch responded with only {:invitation-token token},
so the session start received a map without id or auth-backend. With
the new auth-origin vocabulary that map fails validation and the login
breaks in development and tests; in production the event would be
emitted without a channel.
Return the profile plus the token, like the regular branch, and cover
both branches with RPC-level tests.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* 🐛 Mint an auth token in the OIDC callback with the invitation inside
The callback used the invitation JWE as the session token, so the
frontend landed directly on token verification and never ran the
regular session start: no signin event, no teams, no host hook.
Always mint an :auth token and carry the invitation in its claims.
Token verification returns the claims verbatim, so login-from-token
attaches the invitation to the profile and logged-in does the rest,
the same pattern the verify-email token already uses.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* ✨ Keep the session origin in the telemetry shadow rows
The anonymisation filter drops every string unless it is on a short
allowlist. Add source and activation next to lang and auth-backend so
the shadow copy keeps them; the email and the full name are still
dropped and the email domain is still derived from the email.
The filter keeps its two bindings: the transducer discards everything
that is not a UUID, a boolean or a number, so the allowlist is applied
on the filtered map, never on the raw props.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* ✨ Describe the signup outcome with activation, not with action
The three success branches of register-profile labelled a new account
as "login", "email-verification" or "accept-invitation", so the event
contradicted its own name and the value only lived in context, where
the analytics consumers never read it.
Add activation to the event props and drop action from those branches:
the name already says it is a signup, and a constant action would only
add noise. The retry branches keep their action in context because
there it distinguishes four different outcomes.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* ✨ Describe the session origin in the signin event
The event carried the profile but never said why the session started:
the only field that tried was frontend metadata that never leaves the
browser, and several flows filled it with the wrong value.
Add a small closed vocabulary (source, auth-backend, activation,
from-invitation) and compose it from the profile when signin is
emitted. The assertion fails in development and tests and is elided in
the release build, so a bad value cannot break a real login.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* ✨ Emit the session origin from every login entry point
The four emitters stamped a free-form string as metadata, so the
session event could not rely on it: two flows that were signups wrote
the same value as a login.
Route them through origin-meta: password and LDAP logins declare
source login, register declares source register with activation
immediate, and login-from-token derives both from the token issuer:
:auth for a login and :verify-email for a signup that needed the
email confirmation.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* 🐛 Send the invitation signups through the regular session start
Registering from a team invitation link navigated by hand to token
verification, so the session start never ran: no signin event, no
teams, no host hook. The same navigation was copied in four places.
Teach login-from-register to carry an optional invitation token,
extract the profile decoration into with-register-origin, and call it
from the four sites instead of the hand-written navigation.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* 🐛 Resolve the invitation without waiting for the teams
Registering from an invitation link now goes through the session
start, and that start only navigates after the teams request
completes. fetch-teams has no error handling, so a network failure
there would leave the invitation unaccepted and the user on the signup
screen with a session already minted.
Move the invitation navigation out of the teams path: it emits as soon
as the profile carries a token, and the teams path emits nothing for
an invited profile so it cannot move the user away from the token
verification screen.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* 📚 Document the session origin vocabulary and the invitation flow
Add a focused memory with the four fields, where each value comes from,
the table of flows, the LDAP exception and the invitation path through
logged-in, and link it from the frontend routing.
Update the audit-log memory: source and activation are now on the
telemetry allowlist, activation lives in props and action only exists
on the retry rows.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
* 📎 Cover the invitation session start and fix the auth memories
The invitation flows and the token decoration in login-from-token were
only covered by helper tests, so removing the guard that keeps the
teams path from navigating, or the token attach, broke the behaviour
silently.
Run the events through ptk/watch with a teams-fetched event as input:
an invited profile navigates to the token verification screen and
never to the dashboard, and the signin keeps from-invitation. Add the
login-from-token cases with and without the token, and fix the memory
statements about action, the flow table and from-invitation.
Closes#12849
AI-assisted-by: deepseek-v4.1-flash
A view URL without a page-id crashed builds with assertions enabled:
the viewer passes a nil page-id to `initialize`, whose schema accepted
a missing key but not a nil value. Release builds skip that check and
draw the first page, but every viewer event reads the page from the
route, so zoom to fit and fill, the next-board arrow, Navigate To
interactions, and comment pins did not work.
`initialize` now accepts a nil page-id, `fetch-bundle` makes it
optional, and `bundle-fetched` replaces a route without page-id by one
naming the file's first page. The viewer then initializes again on the
completed route, as it does when the user changes page. A file with no
pages keeps its route.
AI-assisted-by: claude-opus-5-5
Signed-off-by: Álvaro Tejero Cantero <alvorithm@teje.ro>
A rollback reverts the local data with a rebase, which is not a commit,
so the parts of the workspace that follow commits missed it:
- Comment pins moved in the reverted action stayed where they were, on
screen and on the server, and pins on a board the rollback removed
kept pointing to it. They go back to the reverted position, and pins
on a removed board move to the page root.
- Focus mode stayed on with no shape left when the rollback removed the
focused ones, and the next shape drawn turned it on again. It ends
when its last shape goes, and the selection drops removed shapes.
- Component thumbnails kept the rejected look. The thumbnails of the
reverted mains are rendered and uploaded again, and an upload whose
shapes changed during the render is skipped.
- Under WASM, an open text editor reset to the reverted content but the
canvas kept the old layout. It is laid out again.
- A library sync whose save failed showed no update notice again once
the release undid it. The notice is checked again after that
rollback.
Before it repairs a damaged file, the backend keeps a copy of it as a
system snapshot, and the repair notice points to the version history to
find it. The History panel listed that copy among the autosaves, so it
could not be told apart.
The History panel lists the copies kept before a repair as their own
entry per day, "Before automatic repair, <date>", above that day's
autosaves, and the preview of such a copy uses the same label.
The notice of a rejected save always said the same thing. It did not
name the file when the user had moved to another one, gave no hint when
the page held data that was already invalid, and replaced the notice of
a backend repair still on screen, since the store holds one toast.
The backend flags a rejection whose first commit touches pages or
components that failed validation before it (:preexisting-damage). The
rejection notice names a file other than the open one, says when the
data was damaged before, and folds into a repair notice still on
screen; a later repair notice folds a rejection notice in the same way.
The notice says that the latest changes were undone, no longer that the
rest of the work is saved.
The answer to a save lists the changes other sessions stored before it,
such as the backend's repair of a damaged file. When the answer came
before the message with such a change, the saved commit was
acknowledged first, and the late change was then applied on top of the
local edits instead of under them, out of the server's order.
A saved commit of the open file now stays pending until this tab has
received every change the answer lists from other sessions, so a late
change is applied under the local edits. If one does not arrive within
10 seconds, the file reloads.
When a peer restores a version, the file reloads to it. Edits made on
the replaced version stayed queued: their save failed for good and
locked the workspace, and a rollback or a rebase replayed them on the
restored data.
Loading another version forgets the queued, buffered and saved commits
made on the old one. A save of such a commit, in flight or answered, is
dropped instead of failing or being rolled back, and one save carries
the edits of one version only.
While saves fail the workspace turns read-only, but only some tools
checked it. Edits from the asset, color and token panels, and a drag or
a drawing in progress when the lock came, still made commits that
queued behind the failed save.
commit-changes refuses commits while the save-failure lock holds,
except those that are not user edits: the undo of rolled-back commits,
text position data and component touches. Undo and redo do nothing
while locked. When the lock comes, moves, resizes, rotations, drawings
and the drags of flex, guide and grid handles in progress are cancelled
and commit nothing, and the design panel shows the Inspect view.
Reloading the open file (a peer's restore, the fallback of a rollback)
runs initialize-workspace again without leaving the workspace. The
streams the first run started kept running next to the new ones, so a
commit added one undo entry per reload and each notification was
handled twice. The old subscription also sent its unsubscribe after the
new subscribe, which ended the new one on the server.
A new initialize-workspace, text reflow, shape layout or notification
subscription stops the one before it, and a subscription replaced by
one to the same file sends no unsubscribe.
The backend rejects a save after the call that made the changes has
returned, and the workspace then undoes them. A plugin, or the MCP
agent, went on working with shapes that were no longer there and never
learned why.
The rollback event now names the files that lost commits and the pages
and page shapes their changes touched, and plugins get it as the new
`contentrollback` event (`ContentRollback` type) through a subscription
to the store stream, so no rollback is dropped by debouncing.
The MCP plugin keeps the rollbacks it is told about and sends them with
its next task response, failed or not. The server collects them per tool
call and puts a warning first in that tool's response, naming what was
affected; the instructions tell the agent to read the pages again and
redo what is missing.
Closing the v2 text editor with Escape could crash the workspace. The
text options panel renders once more with the editor it saw last, after
closing has disposed it, and reading the style of a disposed editor
throws.
The panel now takes the values from the shape when the editor is gone.
- Ending a drag inside the frame the shapes already belong to sent an
empty reparent save and bumped the revision. The move commits only
when some shape changes parent.
- When a user restored a version, other sessions reloaded the file with
no notice. They now see a dialog naming who restored which version
(by its name, else its date), saying that later changes are no longer
in the file and, when they had changes not saved yet, that those were
lost.
- Exports drawn in the browser (PDF, SVG, and PNG when the WASM renderer
draws it) and duplicating a file from the dashboard started at once,
so an export could show changes a pending save then rejected, and a
copy could miss queued edits. Both wait for the file's pending saves.
A client export warns that it includes unsaved changes; a copy warns
that it misses them.
A rollback can remove or change shapes the UI still refers to.
- The selection drops shapes that are gone and the color picker that
edited them closes.
- The undo history is cut at the reverted action; it stays when no
entry holds a reverted change. Undo and redo commits carry real undo
changes and the history index they moved from, so a reverted Ctrl+Z
puts the change and the index back.
- An open path edition whose shape the rollback changed closes without
saving the editor's stale copy. The v2 text editor loads the reverted
text again; the WASM editor already showed it.
- The Alt measurements look the hovered shape up in the current data.
- Plugin setters, resize() and rotate() on a removed shape throw
"Shape not found", whatever the plugin's throwValidationErrors flag.
- A save that failed on a short platform failure (a lost or refused
database connection, lack of resources, a server shutting down, a
serialization failure or deadlock, a socket or HTTP timeout, a host
that did not resolve, an interruption) answered 500, which stops
autosave. It now answers 503, which the frontend retries. Other I/O
and database errors keep answering 500, timeouts 504.
- A save answered not-found no longer replaces the workspace with the
access page. It is a terminal save error: the workspace stays open
with the failed save notice, report and reload link. When edit rights
are lost, the notice says how many changes could not be saved.
- A library sync marked the library synced three seconds later,
whatever happened to its save. A rejected sync was undone but stayed
marked: copies stayed out of date and the "updates available" notice
never came back. The mark now waits for the save, a rollback of the
sync brings the notice back, and a failed mark (e.g. the user just
became a viewer) is logged instead of leaving the workspace.
- While saves fail the workspace is read-only. The view-mode banner
says editing is disabled because changes can't be saved, with the
Inspect banner's size and bold lead and no Done button. Users can
still switch pages, pan, zoom and inspect. The lock lifts when a save
lands and brings back the mode the user had or picked meanwhile; the
WebGL context lock keeps the workspace read-only if it lasts longer.
The lock lives in app.main.data.persistence.lock.
- A viewer is never locked. A role change that leaves the user a viewer
keeps the edits that lost permission, queued and on screen, and shows
the count notice again. The role change that brings edit rights back
releases them and rolls them back on screen, as a rejected save would.
Any other terminal failure is released by the next role change or file
opened.
The debug faults `io` (socket timeout, 503) and `eof` (500) cover both
sides of the I/O rule.
A rollback reverts whole undo groups, across files and saves, so that
no user action is left half done. Some actions still split:
- An action that commits several times inside one undo transaction
(e.g. applying a token to several shapes) gave each commit its own
group, so a rejection could revert part of it: shapes ended up linked
to a token whose value was never applied. An undo transaction now
opens with a commit group that every commit made inside it takes.
- The commit that undoes saved commits of a rejected action joined an
open transaction's group, so a later rejection there reverted the
undo too. It gets a group of its own.
- Emptying the open transaction during a rollback dropped its group, so
its later commits were no longer one unit. It keeps its group.
- Waiting for a save ended as soon as the queue emptied, while later
commits could still wait in the buffer. Waiters now also wait for an
empty buffer.
A test covers a token action of three commits in one save whose third
commit is rejected.
A save whose changes failed validation was rejected as a whole: autosave
stopped, the edits made after it piled up unsent, and a reload lost
them all, the valid ones included.
The frontend now sends each commit of a save as its own group with its
commit id (:changes-with-metadata). When applying and validating all of
them at once fails, the backend searches by bisection for the first
commit that fails, keeps the commits before it and answers with
:rejected (that commit and every later one). Only the kept changes are
stored, snapshotted and broadcast, so other users never see the rejected
ones. Flat :changes and groups without ids stay all-or-nothing; database
and network errors still fail the whole request. A rejection without an
error type is reported as an internal error.
The frontend rolls the file back to what the server holds: it undoes
every pending commit of that file, applies again the remote changes that
arrived meanwhile, and redoes the kept commits. Undoing only the
rejected commits would write stale values over a peer's later change to
the same attribute. The rejected and later commits leave the queue, the
buffer and the undo history, and a warning that hides after ten seconds
tells the user. The pure bookkeeping lives in
app.main.data.persistence.commits.
The pending notice of a commit the rollback dropped no longer leaves the
save status at pending, which made "Pin this version" hang and fail.
Debug tools drive these cases by hand: debug.commitInvalidChange() makes a
save fail validation, holdSaves() delays the answers, failNextSave() and
failNextRollback() fail them on purpose (backend flag debug-save-faults),
and persistenceState() shows the queue.
A remote commit undoes the pending local commits, applies the remote
changes and replays the local ones, so the local file ends like the
server will: remote first, local last. Only queued commits took part;
the ones still waiting in the persistence buffer (up to ~3s of edits)
stayed under the remote change. When both touched the same attribute,
the remote value won locally but lost on the server.
The buffer now lives in the store ([:persistence :buffer]) instead of
an rx buffer, so the rebase can reach it. A flush merges it per file
into queue entries that keep the list of commits they carry, and the
leave-page guard reads buffer and queue from the store. A buffer left
by a previous session is queued when persistence starts.
At some point the code around this changed, I believe this used to reduce over
file ids. However today it reduces over entire library maps, so it passes a map
instead of an id to `dm/get-in files`. This then hashes the library which
forces it to decode all of its shapes, and then it fails to lookup the data. So
the component/colors/typography counts were always returning 0.
Using reduce-kv fixes this, and we can just get the library data directly with
`(:data library)`.
On the backstage file in penpot-files this brings the time to the telemetry
snapshot down from 3.8s to 1.5s
AI-assisted-by: claude-5.5-sonnet
Co-authored-by: Andrey Antukh <niwi@niwi.nz>