Repository navigation
Cold 9.24 runs: widgetdrop and widgetrestore, Data Widgets filter errors fail smoke, replacement check at triage, rerun guards - #15
Merged
Conversation
…ted RUN outside --work is ignored
Cold WM1 rerun from update: the module steps read as applied ("11 run, 0 committed"), replay skipped what the
log held, and an older replay's empty Account_Overview copy stayed in the menu (Local Users tab empty). step_update
now stops with exit 12 when replay commits follow this rung's convert and names the convert commit to rewind to;
replay-keep.txt accepts them.
A wrapper's own RUN=13 in the environment became the run folder ("missing source mxbuild toolset: 13/tools", exit
3). An inherited RUN now counts only when --work lies inside it, as setup.sh exports it.
Controls: pipeline-update-replay-* (4), pipeline-run-* (3); red 3 FAIL on 1e87d94, green 7/7.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…1 row 11 ports a stranded module's removed calls Cold WM1 rung 2: DeepLink 10.0.10 has no 11 release and its Java calls Core.retrieveXPathQuery and com.mendix.core.conf.RuntimeVersion, both gone in 11. R18 matched the first without saying what replaces it and missed the second. R18 now detects RuntimeVersion and its notes give the replacements (Core.createXPathQuery(xp) .execute(ctx), Core.getRuntimeVersion()); playbooks/10-to-11.md row 11 says to port them as an agent repair or move to page URLs. Both forms compile against the 11.12.7 public-api jar (javac --release 21); a caller that indexes the 4th version part breaks. Controls: replace-java-rtver-catch, replace-java-rtver-ok-pass; red 1 FAIL on 11e58d1, green 2/2, replace + doc group 69/69. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…t (backlog 153) Cold CMM: no UPG_APP_CODE and no state/app-code, so the run log refused its row and harvest and the feedback report skipped. app_code (UPG_APP_CODE, UPG_APP_NAME, the rung's or rung 1's state/app-code) now falls back to app- plus 6 hex of the .mpr file name's hash and writes it to rung 1's state/app-code, which rung 2 shares. The code holds no name, so runlog's customer-name refusal and the public log stay safe. feedback_finish and record_runlog use it. Controls: pipeline-app-code-derived-catch, -kept-pass, -env-pass; red 3 FAIL on 9a0ee03, green 3/3; run log and feedback controls 29/29 with the hook extractions now including app_code. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…er update-widgets dropped its settings (repair row 33) Cold CMM: Administration 4.5.0's Account_Overview Roles filter lost its association mode (baseType ref, refEntity System.UserRoles) when update-widgets ran under DataWidgets 2.2.1; with 3.11.8 the page shows "Unable to get filter store" on 10.24 and 11. widgetdrop.py stops that before it happens; an app where it already happened had no route but Studio Pro, since mxcli cannot write the filter's association mode. bin/repair/widgetrestore.py copies the widget's subtree (saved Type and Object) from an earlier .mpr of the app with fresh ids over the widget of the same name in the same document, lists each setting that comes back, and is a dry run unless --apply. The build's update-widgets then brings the Type to the installed package and keeps every declared key. On the CMM app it restores baseType, refEntity, refCaptionSource, refCaptionExp and fetchOptionsLazy. Controls: widgetrestore-dry-catch, -apply-pass (settings back, rest of the page kept, ids renewed, rerun is a no-op), -missing-refused-pass; red 3 FAIL on 9a0ee03, green 3/3; doc and widget group 28/28. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…filter store") fails the step Cold run of a 9.24 app, 10.24 and 11.12.7: the MaterialHandler dashboard showed "The attribute is not filterable. Please choose a different attribute." in red boxes over the Mixes overview filters, and the smokewalk step PASSed. landedOk counted an .alert-danger box as a widget error only when its text said "widget"; the 3.x filter errors (taken from the installed DataWidgets package) do not. It now also matches "filter store" and "choose a different attribute". An app's own red alert and a validation message still pass. Control smokewalk-landedok-filter-error-catch: red on 9a0ee03, green here; smokewalk group 4/4. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…backlog 155 Cold run of a 9.24 app: a grid column over an association to an external (OData) entity carried a filter. The 2.x filter took it silently; after the DataWidgets update the 3.x filter shows "The attribute is not filterable" on 10.24 and 11. widgets.md section 3 names the case and the app fix; backlog 155 asks for a model check before the build. Documentation only. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…othing in an unattended run) Cold run of a 9.24 app: rung 2 built and walked 11.12.7 with MicroflowTimer 4.0.3 on the mix wizard pages. The widget calls mx.logger, removed in 11.0; the wizard steps never rendered (TypeError reading 'debug'), 8 journeys passed on 10.24 and failed on 11. Row R38 detects exactly this from the .mpk, but replacement_check.py was an agent-only triage item (10), so the unattended proposal said "not measured: replacements.md". replacement_report now runs it against the last rung at triage, and at propose when the report is missing, like dojo_report; HIT rows go to the proposal and a note. Controls pipeline-replacements-run-catch, -unused-pass, -wired-catch: red 3 FAIL on 9a0ee03, green 3/3; pipeline and triage group 145/145. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Owner
Author
|
CI: 5 controls fail on the runner. 4 fail on main too (cockpit-decide-cards-pass, remedy-keptdeps-file-catch, smoke-users-bad-entity-catch, smokewalk-landedok-pass: no bcrypt, no git user.name, a long runner path). The 5th, smokewalk-landedok-filter-error-catch, is new in this PR and fails like smokewalk-landedok-pass: the runner prints none of the landedOk browser lines, while every unit line above them passes. The full suite on a frozen copy of 96cab81 passes locally: 2750 passed, 0 failed, 34 skipped. Making the landedOk browser part run on the runner is a separate fix. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Second set of fixes from the cold runs of 9.24 apps (CMM 9.24.2 → 10.24.28 → 11.12.7, WM1 to 11.12.7). Every fix has a control that fails on the old code and passes on the new.
What changes
replacement_check.pynow runs at triage, and at propose when no report exists. Before, it was an agent-only triage item, so R38 never named a widget callingmx.logger(removed in Mendix 11) in an unattended run. The HIT count goes to the notes and the proposal.bin/repair/widgetrestore.py, repair row 33): puts a pluggable widget back from an earlier commit when update-widgets already dropped its settings. It restores the settings, gives fresh ids and leaves the rest of the page untouched. It runs dry unless--apply.RUNinherited from the caller's environment is ignored when--workis not inside itapp_codederives a neutralapp-<hex>when no code is set (backlog 153)com.mendix.core.conf.RuntimeVersion. Playbook 10-to-11 row 11 ports a stranded module's removed calls.Checks
🤖 Generated with Claude Code