You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This issue tracks the editor regressions that followed the Blockly unforking ("spork") release in April 2026. The most urgent data-loss bugs were hotfixed and deployed in several waves during April, and the backpack corruption repair went live on 2026-06-01. A number of reported problems are still open, some of them for a while, and all of them are kept here.
To make the remaining work easier to follow, and to let each item be picked up and closed on its own, the open items have been split into individual issues, each carrying the spork label and attached to this issue as a sub-issue. This issue stays as the index and the record of what happened. Fixed items and the original detailed triage are preserved further down.
Every open item is now its own issue, attached to this issue as a sub-issue. The Sub-issues panel above is the live status board; the list further down is the same set of issues, grouped by area for easier scanning.
Found a new spork-related bug? Please open a new issue, adding the spork label if you can, instead of commenting here, so it can be tracked and closed on its own.
Deploys reach scratch.mit.edu in waves, so a merged fix and a live fix are different moments. The wave log records when each fix went live.
Live status is shown in the Sub-issues panel above. The list below is the same set of issues, grouped by area for easier scanning. If your report is here and still happening for you on the current site, a comment on that issue confirming it is genuinely helpful.
Statuses below reflect the April to June 2026 triage and may be out of date; the checklist above is the current view. This section is kept for the record and for the analysis links within it.
Expand the full P0–P5 triage (April–June 2026)
P0 -- Data loss / code deletion
✅ Disappearing blocks on project load. Fixed by #3547 + argumentdefaults follow-up.
Cursed inputs from custom block input deletion NOT fixed. Deleting a custom block input still creates a cursed/blank input. Confirmed still reproducible by @nimeratus and @redspacecat (2026-04-22). May share a code path with the "detached input fragments" reports below. Top fix priority.
Variable names mangled (shared "my variable" IDs across projects).
Backpack broadcasts don't define the broadcast variable, so they can never fire (@nimeratus analysis). Independently re-confirmed by pippy2011eight (2026-04-29).
Variable inputs break when importing a backpacked script or sprite (2026-06-11) -- may be related to other reported backpack issues
Detached string input fragments in custom blocks ("potato blocks"). Disconnected string input shapes appear floating in the workspace, can be typed into but not dragged. Reported by fortyonegames; ywc2 notes they "pop out of blocks they originally belonged to, particularly from custom blocks." Likely shares a code path with the custom block input deletion cursed input bug.
Repro confirmed by @kaylen-s: dragging blocks into stacks behind the palette can delete unintended parts of the stack. (2026-04-30)
🤔 Sprite duplication causes code loss. All reports pre-date the 2026-04-21 hotfix; QA suggests this is likely fixed by one of the earlier hotfixes (probably the shared-shadow-ID fix in #3560/#3563). Needs re-verification on the current build.
Old projects: blocks disappear on load and block palette tab disappears + can't be reopened. Reported on ChromeOS Chrome 146. May be aftermath of cursed inputs in saved projects from before the 2026-04-21 hotfix.
RTL editor interactions malfunction: reversed text and operators, erratic drag behavior, and custom blocks can disappear off-screen, making the editor unusable in affected languages.
Received a Freshdesk report from a Hebrew language program with ~100 students who are experiencing significantly disrupted classroom use
P2 -- Blocks editing broken
"Create new broadcast/message" dialog closes immediately.
Sprite costumes skipped when running a tight next costume + wait 0.1 loop over 100+ costumes. Likely runtime/timing rather than spork. https://scratch.mit.edu/discuss/topic/881012/
"Property of sprite" block deletes itself when changing the sprite dropdown. Workaround: edit in the sidebar first.
Undo in costume editor replaces entire sprite with previous sprite. Hitting undo "too many times" in the costume editor reverts so far that the current sprite is replaced by the previously-edited sprite. Single report so far on ChromeOS / Chrome 132.
What this issue tracks
This issue tracks the editor regressions that followed the Blockly unforking ("spork") release in April 2026. The most urgent data-loss bugs were hotfixed and deployed in several waves during April, and the backpack corruption repair went live on 2026-06-01. A number of reported problems are still open, some of them for a while, and all of them are kept here.
To make the remaining work easier to follow, and to let each item be picked up and closed on its own, the open items have been split into individual issues, each carrying the
sporklabel and attached to this issue as a sub-issue. This issue stays as the index and the record of what happened. Fixed items and the original detailed triage are preserved further down.Big thanks to @redspacecat, @nimeratus, @mxmou, @Joeclinton1, @npxl32, and others for collecting reports and contributing analysis and fixes. The community-maintained list is at https://scratch.mit.edu/discuss/post/9109627/.
How the list works now
sporklabel: https://github.com/search?q=org%3Ascratchfoundation+label%3Aspork+state%3Aopen&type=issuessporklabel if you can, instead of commenting here, so it can be tracked and closed on its own.Deploy waves (live on scratch.mit.edu)
Open items
Live status is shown in the Sub-issues panel above. The list below is the same set of issues, grouped by area for easier scanning. If your report is here and still happening for you on the current site, a comment on that issue confirming it is genuinely helpful.
Data loss:
Broken behavior:
Editing and interaction:
Visual:
Performance:
Believed fixed, please re-check
These are believed resolved by earlier fixes but not yet confirmed on the live site. If any still happens for you, please say so on the issue.
Archived history
Statuses below reflect the April to June 2026 triage and may be out of date; the checklist above is the current view. This section is kept for the record and for the analysis links within it.
Expand the full P0–P5 triage (April–June 2026)
P0 -- Data loss / code deletion
stripIdsmutated original block'sconnection.shadowStateby reference, creating orphaned top-level shadows that prevent workspace rendering on sprite switch. Diagnosis from @nimeratus, clear repro from @Joeclinton1.P1 -- Silent wrong behavior / runtime failures / interaction failures
P2 -- Blocks editing broken
P3 -- Visual / cosmetic regressions
box-sizing: border-boxmissing in production. Fix identified by @mxmou.P4 -- Performance / viewport
P5 -- Low-confidence / adjacent / unreproduced
next costume+wait 0.1loop over 100+ costumes. Likely runtime/timing rather than spork. https://scratch.mit.edu/discuss/topic/881012/