docs/slate-issues/open-issues-dossiers/4963-4842.md
LeeKatz1bug1Autocorrect on iOS can revert to the original misspelled word as soon as the user presses space, which makes Slate fight the platform correction flow.
The only follow-up says this was fixed later, so this looks more like release drift than live engine debt.
Strong enough.
Acceptable.
Stale candidate.
This likely belongs in stale-history cleanup, not active architecture work.
close-stale
Close it as stale history unless someone can still reproduce it on a current release.
Indirect.
Not a direct test candidate.
gaoyuan1223mbug, ⚑ cross platform1QQ Browser plus Sogou input cannot reliably compose Chinese text in Slate at all.
The thread is thin, but the report is specific enough to treat this as browser-and-IME compatibility pressure rather than random support noise.
Weak.
Poor.
Likely valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Indirect.
Blocked on a tighter repro.
sodennbug9ReactEditor.focus can land the caret in the wrong place after a new node is inserted, especially in flows like mentions or post-insert cursor moves.
The thread is full of workarounds and confirmations, which makes this a real runtime timing bug rather than one bad call site.
Strong enough.
Acceptable.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Direct.
Ready now.
ReedRodgersbug, ⚑ cross platform2Holding backspace on Android stalls at block boundaries, skips deletions, and can dismiss the keyboard mid-edit.
Follow-ups connect it to the broader Android fast-input mess, so this is part of a real recurring cluster.
Strong enough.
Poor.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Direct.
Ready with minor setup.
goowiknsimprovement7The examples are useful as demos but too many of them fall apart when copied into a real strict TypeScript project.
The thread adds multiple confirmations from new users, so this is real docs and example friction, not one picky tsconfig complaint.
Strong enough.
Acceptable.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
None.
Not a direct test candidate.
cameronbraidbug1A regression changed triple-click paragraph selection from one block to multiple blocks.
The repro is sharp and version-scoped, which makes this feel like a real selection regression rather than UX preference noise.
Strong enough.
Poor.
Likely valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Indirect.
Ready now.
wfischer42improvement1This asks for a way to force useEffect instead of useLayoutEffect during HTML serialization paths that call ReactDOMServer from the client.
The issue itself admits this is mostly not Slate’s problem, so it should not masquerade as core engine debt.
Strong enough.
Acceptable.
Likely invalid.
This is mostly unsupported or invalid current-contract behavior.
close-invalid
Close it as unsupported or out-of-contract behavior instead of pretending it is core debt.
Indirect.
Not a direct test candidate.
bknillbug0The docs deserialization snippet recurses on the wrong variable and immediately blows the stack for simple HTML.
This is the same underlying docs bug as #4882, just reported from a user hitting the broken sample code.
Strong enough.
Poor.
Valid.
This is real issue signal, but it should point at #4882 instead of floating separately.
mark-duplicate
Point it at the canonical issue instead of letting duplicate tracker weight inflate the theme.
None.
Not a direct test candidate.
juliankrispelbug0wrapNodes with split: true and match can produce structurally wrong results around overlapping inline content.
The report is detailed and concrete enough that this belongs in current transform correctness work, not just v2 wishlists.
Strong enough.
Poor.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Direct.
Ready now.
jonahallibonebug18The TypeScript story around node.type is still awkward enough that users fall into casts, doc workarounds, and any just to toggle formatting.
The thread is long and recurring, which makes this a real typing/API design problem rather than a one-off docs confusion.
Strong enough.
Acceptable.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Indirect.
Not a direct test candidate.
jasontlourobug2Copying through Slate can serialize a single blank line as two blank lines when pasted into plain text.
The report is clean, but the small thread means this still needs a focused repro lane instead of overclaiming a root cause.
Strong enough.
Acceptable.
Likely valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Indirect.
Ready with minor setup.
mwood23improvement0This is straightforward repo DX work: more reusable test helpers would make slate-react regressions easier to catch.
There is no real thread, but the request is clear and bounded.
Strong enough.
Acceptable.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
share-status
Reply with the current contract and the right ownership boundary instead of overpromising a fix.
None.
Not a direct test candidate.
cmmartinbug3Deleting an empty text node after a void can also delete the void itself, which makes voids unstable at document boundaries.
There is a workaround via normalization, but that is exactly why this belongs in the core behavior bucket.
Strong enough.
Acceptable.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Direct.
Ready now.
ulionnone1The Editable copy path behaves differently in readOnly, and the question is really about what the contract is supposed to be.
The only follow-up immediately points at #4914, so this should not float separately.
Strong enough.
Poor.
Likely valid.
This is real issue signal, but it should point at #4914 instead of floating separately.
mark-duplicate
Point it at the canonical issue instead of letting duplicate tracker weight inflate the theme.
Indirect.
Not a direct test candidate.
stevemarksdbug1Dropping inside a void still routes through Slate’s normal drop handling and triggers insertData, even when the void wants to own the drop.
The workaround is explicit and the ownership seam is crisp, so this is a good runtime bug, not support mush.
Strong enough.
Acceptable.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Direct.
Ready now.
ohansemmanuelbug, ⚑ cross platform0The docs point at a broken HTML deserialization implementation that throws on basic input.
This is a straightforward docs bug with a suggested fix already in the issue.
Strong enough.
Strong.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
None.
Not a direct test candidate.
AxelBrichebug, ⚑ cross platform0Triple-clicking a paragraph and then moving the cursor up loses the initial selection instead of creating the expected block selection.
The repro is on the official examples, so this is a real selection bridge problem.
Strong enough.
Poor.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Indirect.
Ready now.
alex-friedbug, ⚑ cross platform0On Android, changing block type after typing can delete the line entirely.
The report is short but concrete and directly tied to the official richtext example.
Strong enough.
Poor.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Direct.
Ready with minor setup.
stevemarksdbug5Select-all and paste in the HTML example can still throw null-path style errors at fragment boundaries.
The thread has real debugging, prior PR context, and partial fixes, so this is strong current bug signal.
Strong enough.
Acceptable.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Indirect.
Ready with minor setup.
rxxndyimprovement3The complaint is real enough to recognize, but the issue is still too vague to be a strong architecture input on its own.
The comments basically say “we see this too,” which is useful cluster signal but not enough issue-level precision.
Weak.
Poor.
Unclear.
The report is not sharp enough yet to carry strong architecture weight.
ask-for-repro
Ask for a tighter reduced repro before letting it influence architecture or roadmap work.
Indirect.
Blocked on a tighter repro.
antondcbug1Using KaTeX-rendered math inside the editor can break cursor movement and blow up DOM-point resolution.
The repro is real, but ownership is muddy because third-party rendered DOM can easily violate Slate’s assumptions.
Weak.
Poor.
Unclear.
The report is not sharp enough yet to carry strong architecture weight.
share-status
Reply with the current contract and the right ownership boundary instead of overpromising a fix.
Indirect.
Blocked on a tighter repro.
aliechtibug, ⚑ cross platform0This is really a request to document the browser support floor more clearly when people try to transpile Slate into old-browser support.
There is no thread, and the issue is explicitly about unsupported old-browser behavior.
Strong enough.
Acceptable.
Stale candidate.
This likely belongs in stale-history cleanup, not active architecture work.
close-stale
Close it as stale history unless someone can still reproduce it on a current release.
None.
Not a direct test candidate.
M162bug, ⚑ cross platform5Firefox Chinese input can duplicate text and then crash the editor, but the thread also points to earlier duplicates and a claim that latest releases fixed it.
This should stay in the IME recurrence signal, but not pretend to be fresh unresolved debt without a current repro.
Strong enough.
Acceptable.
Stale candidate.
This likely belongs in stale-history cleanup, not active architecture work.
close-stale
Close it as stale history unless someone can still reproduce it on a current release.
Indirect.
Not a direct test candidate.
BashmundShahbug, ⚑ cross platform0Deleting content in Safari can shove the whole DOM upward instead of keeping the caret in view sanely.
The official example repro makes this a real Safari bridge bug, not app-specific CSS weirdness.
Strong enough.
Poor.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Indirect.
Ready with minor setup.
JosNunbug0Nested editors can produce the wrong toSlatePoint offset while the mouse is down, and the bogus selection can leak into history.
This is a great current-contract bug: precise repro, cross-package impact, and clear invalid-state leakage.
Strong enough.
Poor.
Valid.
Keep it in the right cluster and do not let it drift into unrelated themes.
keep-open
Keep it open and scoped to the actual failing seam instead of letting it dissolve into vague frustration.
Direct.
Ready now.