full_git_push: el auto-recover de non-fast-forward nunca se disparaba por dos
bugs. (1) git_push_if_ahead hacia `tail -1` del error de push y se quedaba con
la linea final de `hint:`, perdiendo "[rejected]" y "Updates were rejected", de
modo que el pipeline no detectaba el rechazo. Ahora se preservan las lineas con
los keywords (rejected/fast-forward/fetch first). (2) El paso 4 capturaba stdout
y stderr juntos (2>&1), metiendo lineas "[push]" de stderr al principio del
status y rompiendo el glob `== "[error]"*` anclado al inicio; ahora solo captura
stdout y la deteccion cubre todas las redacciones de git. (3) El merge auto
evaluaba el exit de `tail` (siempre 0) en vez del de `git merge`; un merge con
conflictos se reportaba como exito. Ahora se evalua el exit real del merge y se
aborta limpio ante conflictos.
Limpieza: cpp/apps/chart_demo y cpp/apps/shaders_lab estaban commiteados como
gitlinks (submodulos, mode 160000) sin URL en .gitmodules, directorios vacios
fantasma que rompian el paso 3 de full_git_pull con "fatal: No url found for
submodule path". Eliminados del indice. Las apps C++ reales viven en apps/ como
sub-repos Gitea independientes (regla cpp_apps.md; cpp/apps/ deprecado por
issue 0096).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Auto-init step skipped any dir with .git present, even when it lacked
an origin remote. Such dirs fell through to push step and failed with
"'origin' does not appear to be a git repository". Now skip only when
.git AND origin exist; otherwise run ensure_repo_synced to create the
Gitea repo + add origin + push.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Foundation (ce7470d5) + reader rewire (2f7fdd40).
- ColumnSnapshot per col (i64/f64/str_ids) + StringPool per-State
- compute_visible_rows filter/sort uses snapshot direct numeric/id compare
- StringPool realloc-crash fix (reserve before emplace_back)
- Pool staleness sentinel (rebuild when string_pool.size() drift)
- High-cardinality cap (>2048 unique → skip interning, fallback raw)
API publica intacta. Bench 100k sort_numeric +131% vs baseline.
text_editor_smoke RED preexisting unrelated.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Change 3 of issue 0133 — rewire compute_visible_rows, filter eval,
and sort comparators to read from the SnapshotCache when available.
Hot paths rewired:
- compute_visible_rows (overload with snap): filter eval uses
compare_snap (fast i64/f64 numeric compare for Int/Float cols;
id-compare for low-cardinality string Eq/Neq; raw cells fallback
for Contains/StartsWith/EndsWith).
- Sort comparators: direct i64/f64 array compare for Int/Float cols
(goto sort_done skips string fallback); string sort uses uint32_t
id compare with pool lookup only on mismatch.
- Stage>0 filter/sort: same snapshot overload.
Materialization paths (build_so, s0_backing, mat_backing, config popup)
kept on raw cells — they copy into std::string anyway, no benefit from
snapshot and snprintf-per-cell was 2M extra calls per frame.
Bug fixes (required for correctness):
1. StringPool::intern() realloc safety: force reserve before
emplace_back so string_view keys in the map never go dangling.
2. SnapshotCache::pool_size_built sentinel: detects when a new State
is created with an empty pool but same cells pointer (begin_scenario
pattern). Prevents str_ids from indexing into an empty pool (SIGSEGV).
3. Cardinality cap (2048 uniques / 25% sample): high-cardinality string
cols (timestamps-as-strings, UUIDs, names) skip interning — str_ids
stays empty and compare_snap falls back to raw cells. Prevents 30MB+
pool bloat that hurt cache for filter/sort on other cols.
Bench delta vs baseline (100k rows, LIBGL_ALWAYS_SOFTWARE=1):
linear_scroll: 16.0 -> 15.5 fps p50 (-3%, baseline already FAIL)
filter_like: 59.7 -> 56.0 fps p50 (-6%, still PASS at 56fps)
sort_numeric: 3.9 -> 9.0 fps p50 (+131%, snapshot i64 sort)
color_rule: 15.2 -> 14.8 fps p50 (-3%, baseline already FAIL)
Build: green for all 10 available Linux consumers (text_editor_smoke
linker failure is preexisting, not caused by this change).
API public intact. TableEvent.row indexing TableInput preserved.
Pointer-identity invalidation preserved.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
terminal_panel.cpp:
- BeginChild con PushStyleColor(ChildBg, negro) + PushStyleColor(Text, gris claro)
- PushStyleVar(WindowPadding, 8/6px) para padding terminal real
- Input prompt siempre visible cuando readonly=false
- Prefijo "$ " antes del InputText (TextUnformatted + SameLine)
- BeginDisabled() cuando el shell esta cerrado (en vez de ocultar el widget)
- Calculo de child_h reserva exactamente GetFrameHeightWithSpacing+6 para el prompt
cpp/tests/e2e/test_terminal_panel_e2e.py (nuevo):
- 4 asserts: PNG existe, no todo-blanco, region oscura >= 30%, pixels no-negros >= 0.3%
- Lanza primitives_gallery --capture, busca el binario Linux o Windows.exe automaticamente
- Skip graceful si no hay GL ni binario (WSL/CI headless)
- 4/4 pasan en Linux con LIBGL_ALWAYS_SOFTWARE=1
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds the add_subdirectory block for apps/data_table_bench so the build
system picks it up. The app itself lives in its own sub-repo.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Change 1 — Columnar Snapshot Internal:
- Add ColumnSnapshot struct (type + str_ids/i64/f64 per column) in data_table_internal.h
- Add SnapshotCache struct with pointer-identity sentinel (last_cells_ptr)
- Add SnapshotCache field to UiState singleton
- In render(): rebuild snapshot after join materialization when cells ptr changes
Uses same pointer-identity pattern as existing stats_last_cells in State
Int/Float columns parsed once via parse_number; String/Auto interned
Change 2 — String Interning:
- Add StringPool struct (strings + unordered_map<string_view, uint32_t>) to data_table_types.h
- StringPool is per-State (NOT global) for table isolation
- intern(sv) inserts if absent, returns stable uint32_t index
- Cleared + rebuilt on each snapshot rebuild for index coherence
- Add string_pool field to State struct
Documentation:
- Extended header comment in data_table_internal.h describing design,
StringPool API, invariants (pointer-identity, row→snapshot_row),
and how stats_last_cells and snapshot coexist independently
Build: fn_module_data_table + tables_qa pass, no new errors (only
pre-existing -Wformat-truncation warnings unrelated to this change).
Public API (data_table.h, TableInput, render() signature) unchanged.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
POSIX popen routes via /bin/sh -c, so "2>&1" is a shell redirect. On
Windows we use CreateProcessW directly (no shell): curl receives "2>&1"
as a positional arg, treats it as a second URL, and fails with exit 3
"URL rejected: Bad hostname".
Stderr is already merged into the same pipe via STARTUPINFOW.hStdError
on Windows, so the redirect is also unnecessary there. Guard with
#ifndef _WIN32.
Also adds FN_HTTP_DEBUG env var to dump the cmdline + req.url for
future bug triage (zero-cost when unset).
Detected via agents_dashboard.exe --connect-test against
https://agents.organic-machine.com — same .exe with the fix now returns
"OK 11" in <2s.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Both issues delivered end-to-end:
0128 (backend, merged via dataforge/agents_and_robots/pulls/1):
- HTTP daemon in cmd/launcher with apikey Bearer auth + SSE
- LIVE at https://agents.organic-machine.com via Coolify Traefik + LE cert
- systemd Restart=always
- Unified status autodetect fix applied
0129 (frontend, merged via dataforge/agents_dashboard/pulls/1):
- C++ ImGui app in projects/element_agents/apps/agents_dashboard
- 4 panels: Connection / Agents / Logs / Status
- secret_store_cpp_infra new function (DPAPI Windows / XOR Linux)
- Deployed to Windows Desktop, App Hub tarjeta visible
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
DPAPI Windows + XOR Linux fallback para almacenar credentials sensibles
en SQLite local. Usado por agents_dashboard para cifrar apikeys.
Incluye encrypt/decrypt/is_strong + base64 helpers.
Issue: 0129
Co-Authored-By: fn-constructor <noreply@fn-registry.local>
Gitea repo dataforge/kanban_cpp archived (read-only).
Local apps/kanban_cpp/ deleted, CMake subdir registration removed.
registry.db entry + pc_locations row purged (regenerable via fn index +
manual delete since indexer upserts but does not purge orphaned apps).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Doble entrada confusa (incidente 2026-05-19 piloto 0121b: cwd
mutation por Path B inline causo commit a branch incorrecta).
Cambios:
- .claude/commands/autopilot.md: v2 simplificado. SOLO pre-flight
DoD check + delegate fn-orquestador. Sin Path A/B/C inline.
Self-Q&A migrado al orquestador. Cero cwd mutation.
- .claude/commands/autonomous-task.md: DEPRECADO. Sustitucion 1:1.
Sigue funcionando como debug primitive sin DoD check.
- dev/issues/0123: revision — eliminar /flow run y /fix-flow (absorbidos
por /autopilot v2). Mantener fn-meta-orquestador + fn-priorizador
+ fn doctor issues/flows. Anadir tarea: dar a fn-orquestador soporte
task_type=flow.
Preferencia humano: 1 sola entrada autopilot, "modo que entra y sigue".
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
task_run task_d285372493cce2e6 converged 1 iter / ~4 min.
PR https://gitea-.../dataforge/fn_registry/pulls/3 mergeado.
Verificado en master:
total=45 with_checks=19 coverage=42.22%
21 apps con propuesta lista en dev/proposals_e2e_checks_0121/ esperando aplicacion (0121c).
5 apps sin propuesta aun (wave 4 pendiente).
Desbloquea: 0121c (apply N PRs add_e2e_check).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Crea functions/infra/audit_e2e_coverage.go: AuditE2ECoverage(roots) escanea
app.md recursivamente, detecta e2e_checks: en frontmatter, retorna
E2ECoverageReport{total, with_checks, missing, coverage_pct}.
- Crea functions/infra/e2e_coverage_report.go: tipo E2ECoverageReport con
JSON tags (total, with_checks, missing, coverage_pct).
- Crea types/infra/e2e_coverage_report.md: metadata del tipo para registry.
- Crea functions/infra/audit_e2e_coverage.md: documentacion self-contained
con Ejemplo, Cuando usarla, Gotchas.
- Crea functions/infra/audit_e2e_coverage_test.go: 3 tests (empty, all-covered,
partial) — todos pasan.
- Edita cmd/fn/doctor.go: agrega case "e2e-coverage" -> doctorE2ECoverage().
Output text (tabla tabwriter + lista de apps missing) y --json (E2ECoverageReport).
Acceptance verificado:
fn doctor e2e-coverage --json -> {total, with_checks, missing, coverage_pct} OK
fn doctor e2e-coverage -> tabla text OK
go test ./functions/infra/... -> 3/3 PASS
fn show audit_e2e_coverage_go_infra -> indexada OK
task_run: task_d285372493cce2e6 iter 1
Co-authored-by: fn-orquestador <noreply@fn-registry>
Origen: fn-recopilador design-e2e descubrio 6 bugs durante el design
de propuestas e2e_checks. Agrupados en 4 issues:
- 0124 EPIC dag_engine cleanup (registry.db huerfana + Mantine drift
+ --migrate-only flag — 3 sub-tareas)
- 0125 deploy_server: anadir --db a cmdServe
- 0126 pipeline_launcher: aplicar migracion 003_logs
- 0127 docker_tui: go.work path absoluto rompe portabilidad
Todos relacionados con 0121a. Pueden ser candidatos a /autonomous-task
o /autopilot dependiendo del scope.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
0121 era scope demasiado grande para 1 orquestador (batch + new function +
new subcmd + N edits). Split:
- 0121a chore: design-e2e batch (Claude orquesta N fn-recopilador paralelos)
- 0121b feature: audit_e2e_coverage + fn doctor e2e-coverage subcmd
(tipo feature_app_simple, apto /autonomous-task)
- 0121c chore: aplicar propuestas via N /autonomous-task add_e2e_check
Cada hijo tiene Acceptance verificable + tipo orquestador declarado.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
egutierrez
egutierrez
dataforge