Guard imperative DOM operations (removeChild, RAF callbacks) against
race conditions where React has already unmounted the target nodes.
- assets/common.cljs: add dom/child? guard before removeChild in RAF
- dynamic_modifiers.cljs: capture RAF IDs and cancel them on cleanup;
add null guards for DOM nodes that may no longer exist
- hooks.cljs: guard portal container removal with dom/child? check
- errors.cljs: extract is-ignorable-exception? to a top-level defn
and add NotFoundError/removeChild to ignorable exceptions, since
these are caused by browser extensions modifying React-managed DOM
- Add unit tests for is-ignorable-exception? predicate
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
* 🚑 Fix RangeError from re-entrant error handling in errors.cljs
Two complementary changes to prevent 'RangeError: Maximum call stack
size exceeded' when an error fires while the potok store error pipeline
is still on the call stack:
1. Re-entrancy guard on on-error: a volatile flag (handling-error?)
is set true for the duration of each on-error invocation. Any
nested call (e.g. from a notification emit that itself throws) is
suppressed with a console.error instead of recursing indefinitely.
2. Async notification in flash: the st/emit!(ntf/show ...) call is
now wrapped in ts/schedule (setTimeout 0) so the notification event
is pushed to the store on the next event-loop tick, outside the
error-handler call stack. This matches the pattern already used by
the :worker-error, :svg-parser and :comment-error handlers.
* 🐛 Add unit tests for app.main.errors
Test coverage for the error-handling module:
- stale-asset-error?: 6 cases covering keyword-constant and
protocol-dispatch mismatch signatures, plus negative cases
- exception->error-data: plain JS Error, ex-info with/without :hint
- on-error dispatch: map errors routed via ptk/handle-error, JS
exceptions wrapped into error-data before dispatch
- Re-entrancy guard: verifies that a second on-error call issued
from within a handle-error method is suppressed (exactly one
handler invocation)
---------
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
Clojure's = uses .equals on doubles, and Double.equals(Double.NaN)
returns true, so (not= v v) was always false for NaN. Use
Double/isNaN with a number? guard instead.
The CLJS branch of num-string? checked (string? v) first, but the
JVM branch did not. Passing non-string values (nil, keywords, etc.)
would rely on exception handling inside parse-double for control
flow. Add the string? check for consistency and to avoid using
exceptions for normal control flow.
The removev function used 'fn' as its predicate parameter name,
which shadows clojure.core/fn. Rename to 'pred' for clarity and
to follow the naming convention used elsewhere in the namespace.
When called with an empty string as the base class, append-class
was producing " bar" (with a leading space) because (some? "")
returns true. Use (seq class) instead to treat both nil and empty
string as absent, avoiding invalid CSS class strings with leading
whitespace.
The :else branch of diff-attr was calling (get m1 key) and
(get m2 key) again, but v1 and v2 were already bound to those
exact values. Reuse the existing bindings to avoid the extra
lookups.
The docstring claimed the function removes nil values in addition to
the specified object, but the implementation only removes elements
equal to the given object. Fix the docstring in both data.cljc and
the local copy in files/changes.cljc.
The 3-arity of safe-subvec called (count v) in a let binding before
checking (some? v). While (count nil) returns 0 in Clojure and does
not crash, the nil guard was dead code. Restructure to check (some? v)
first with an outer when, then compute size inside the guarded block.
Replace (empty? items) + (rest items) with (seq items) + (next items)
in enumerate. The seq/next pattern is idiomatic Clojure and avoids
the overhead of empty? which internally calls seq and then negates.
The index-of-pred function used (nil? c) to detect end-of-collection,
which caused premature termination when the collection contained nil
values. Rewrite using (seq coll) / (next s) pattern to correctly
distinguish between nil elements and end-of-sequence.
The deep-mapm function was applying the mapping function twice on
leaf entries (non-map, non-vector values): once when destructuring
the entry, and again on the already-transformed result in the else
branch. Now mfn is applied exactly once per entry.
The patch-object function was calling (dissoc object key value) when
handling nil values. Since dissoc treats each argument after the map
as a key to remove, this was also removing nil as a key from the map.
The correct call is (dissoc object key).
Tests that exercise app.common.types.color were living inside
common-tests.colors-test alongside the app.common.colors tests. Move
them to common-tests.types.color-test so the test namespace mirrors
the source namespace structure, consistent with the rest of the
types/ test suite.
The [app.common.types.color :as colors] require is removed from
colors_test.cljc; the new file is registered in runner.cljc.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
The five functions interpolate-color, offset-spread, uniform-spread?,
uniform-spread, and interpolate-gradient duplicated the canonical
implementations in app.common.types.color. The copies in colors.cljc
also contained two bugs: a division-by-zero in offset-spread when
num=1, and a crash on nil idx in interpolate-gradient.
All production callers already use app.common.types.color. The
duplicate tests that exercised the old copies are removed; their
coverage is absorbed into expanded tests under the types-* suite,
including a new nil-idx guard test and a single-stop no-crash test.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
Add indexed-access-with-default in fill_test.cljc to cover the two-arity
(nth fills i default) form on both valid and out-of-range indices, directly
exercising the CLJS Fills -nth path fixed in 593cf125.
Add segment-content->selrect-multi-line in path_data_test.cljc to cover
content->selrect on a subpath with multiple consecutive line-to commands
where move-p diverges from from-p, confirming the bounding box matches
both the expected coordinates and the reference implementation; this
guards the calculate-extremities fix in bb5a04c7.
Add types-uniform-spread? in colors_test.cljc to cover
app.common.types.color/uniform-spread?, which had no dedicated tests.
Exercises the uniform case (via uniform-spread), the two-stop edge case,
wrong-offset detection, and wrong-color detection.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
In the :line-to branch of calculate-extremities, move-p (the subpath
start point) was being added to the extremities set instead of from-p
(the actual previous point). For all line segments beyond the first one
in a subpath this produced an incorrect bounding-box start point.
The :curve-to branch correctly used from-p; align :line-to to match.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
In the ClojureScript Fills deftype, the two-arity -nth implementation
called (d/in-range? i size) but the signature is (d/in-range? size i).
This meant -nth always fell through to the default value for any valid
index when called with an explicit default, since i < size is the
condition but the args were swapped.
The no-default -nth sibling on line 378 and both CLJ nth impls on
lines 286 and 291 had the correct argument order.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
In the CLJS -lookup implementation, when a key is absent from data the
negative cache entry was stored under 'key' (the built-in map-entry
key function) rather than the 'k' parameter. As a result every
subsequent lookup of any missing key bypassed the cache and repeated
the full lookup path, making the negative-cache optimization entirely
ineffective.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
`(cfh/frame-shape? current-id)` passes a UUID to the single-arity
overload of `frame-shape?`, which expects a shape map; it always
returns false. Fix by passing `current` (the resolved shape) instead.
Update the test to assert the correct behaviour.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
`(mapcat collect-main-shapes children objects)` passes `objects` as a
second parallel collection instead of threading it as the second
argument to `collect-main-shapes` for each child. Fix by using an
anonymous fn: `(mapcat #(collect-main-shapes % objects) children)`.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
\`(get "type" shadow)\` always returns nil because the map and key
arguments were swapped. The correct call is \`(get shadow "type")\`,
which allows the legacy innerShadow detection to work correctly.
Update the test expectation accordingly.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
font-weight-keys was listed twice in the set/union call for
typography-keys, a copy-paste error. The duplicate entry has no
functional effect (sets deduplicate), but it is misleading and
suggests a missing key such as font-style-keys in its place.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
get-children-rec passed the original children vector to each recursive
call instead of the updated one that already includes the current
shape. This caused descendant results to be accumulated from the wrong
starting point, losing intermediate shapes. Pass children' (which
includes the current shape) into every recursive call.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
When no gradient stop satisfies (<= offset (:offset %)),
d/index-of-pred returns nil. The previous code called (dec nil) in
the start binding before the nil check, throwing a
NullPointerException/ClassCastException. Guard the start binding with
a cond that handles nil before attempting dec.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
The key :podition was used instead of :position when updating the
id-from cell in swap-shapes, silently discarding the position value
and leaving the cell's :position as nil after every swap.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
The 'Move to' menu in the dashboard file context menu only filtered
out the first selected file's project from the available target list.
When multiple files from different projects were selected, the other
files' projects still appeared as valid targets, causing a 400
'cant-move-to-same-project' backend error.
Now all selected files' project IDs are collected and excluded from
the available target projects.
lambdaisland/uri's query-string->map uses :multikeys :duplicates by
default: a key that appears once yields a plain string, but the same
key repeated yields a vector. cljs.core/parse-long only accepts
strings and therefore threw "Expected string, got: object" whenever
a URL contained a duplicate 'index' parameter.
Add rt/get-query-param to app.main.router. The helper returns the
scalar value of a query param key, taking the last element when the
value is a sequential (i.e. the key was repeated). Use it at every
call site that feeds a query-param value into parse-long, in both
app.main.ui (page*) and app.main.data.viewer.
Refactor use-portal-container to allocate one persistent <div> per
logical category (:modal, :popup, :tooltip, :default) instead of
creating a new div for every component instance. This keeps the DOM
clean with at most four fixed portal containers and eliminates the
arbitrary growth of empty <div> elements on document.body while
preserving the removeChild race condition fix.
Remove unrelated local pid file that was accidentally included in previous commit.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: raguirref <ricardoaguirredelafuente@gmail.com>
Fixes three concrete builder issues in common/files/builder:\n- Use bool type from shape when selecting style source for difference bools\n- Persist :strokes correctly (fix typo :stroks)\n- Validate add-file-media params after assigning default id\n\nAlso adds regression tests in common-tests.files-builder-test and registers them in runner.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: raguirref <ricardoaguirredelafuente@gmail.com>
