dataforge/agents_dashboard
1
0
Fork 0
Code Issues Pull Requests 2 Actions Packages Projects Releases Wiki Activity

fix(crash): init data_table State.col_visible + col_order before render

Browse Source 
Pressing Refresh Agents (or Test Connection — both trigger fetch + table
re-render) crashed the app with Windows exit code 5 (access violation).

Root cause: agents_tbl_state was default-constructed, so
State.col_visible (std::vector<bool>) and State.col_order
(std::vector<int>) were empty. render_grid_stage0 indexes them by column
index up to N_COLS=11 without bounds checking → undefined behaviour →
segfault on the first render after agents data populated.

Fix: at first render of the agents panel, assign col_visible=true * N_COLS,
fill col_order with [0..N_COLS), and ensure stages.size() >= 1. Same
pattern tql_apply.cpp uses (col_visible.assign(eff_cols, true)).

Diagnostic infra added (kept in place — minimal overhead):
- FN_DBG macro: fprintf(stderr, ...) + fflush. Survives crashes that
  fn_log's buffered file output doesn't.
- --auto-refresh CLI flag: triggers fetch_agents_async at frame 30,
  auto-exits at frame 180 (~3s @ 60Hz). Headless smoke for CI.
- DBG breadcrumbs through main → load_apikey → fn::run_app → render →
  fetch_agents_async (thread enter/request/response/parse/exit) → render
  table (pre/post). Each step flushes stderr immediately.

E2E regression guard: test_app_survives_auto_refresh_cycle. Runs the .exe
with --auto-refresh, asserts exit 0, asserts the breadcrumb chain reaches
both "fetch thread parsed" and "agents_panel POST-render" in stderr. 25
tests passing.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Egutierrez
2026-05-22 23:31:25 +02:00
parent 18b5ffdfd9
commit 9cade2f2f8
2 changed files with 126 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
main.cpp
+91 -16
View File
@@ -373,29 +373,47 @@ static std::vector<AgentRow> parse_agents(const std::string& body) {
return rows;
}
// FN_DBG: stderr + flush. Survives crashes (fn_log buffers).
#define FN_DBG(...) do { fprintf(stderr, "[DBG] " __VA_ARGS__); fputc('\n', stderr); fflush(stderr); } while(0)
// Fetch agents in background thread
static void fetch_agents_async(AppState& s) {
if (s.fetching) return;
FN_DBG("fetch_agents_async ENTER s.fetching=%d apikey_len=%zu",
(int)s.fetching, strlen(s.apikey_buf));
if (s.fetching) {
FN_DBG("fetch_agents_async SKIP already fetching");
return;
}
s.fetching = true;
std::thread([&s]() {
FN_DBG("fetch thread STARTED");
fn_http::Request req;
req.method = "GET";
req.url = make_url(s, "/agents");
req.bearer_token = s.apikey_buf;
req.timeout_ms = 8000;
FN_DBG("fetch thread requesting url=%s bearer_len=%zu", req.url.c_str(), req.bearer_token.size());
auto res = fn_http::request(req);
std::lock_guard<std::mutex> lk(s.agents_mu);
if (!res.error.empty()) {
s.agents_error = "Transport error: " + res.error;
} else if (res.status != 200) {
s.agents_error = "HTTP " + std::to_string(res.status);
} else {
s.agents = parse_agents(res.body);
s.agents_error.clear();
s.agents_fetched_ms = now_ms();
FN_DBG("fetch thread response status=%d err=[%s] body_len=%zu",
res.status, res.error.c_str(), res.body.size());
{
std::lock_guard<std::mutex> lk(s.agents_mu);
if (!res.error.empty()) {
s.agents_error = "Transport error: " + res.error;
} else if (res.status != 200) {
s.agents_error = "HTTP " + std::to_string(res.status);
} else {
FN_DBG("fetch thread parsing body...");
s.agents = parse_agents(res.body);
s.agents_error.clear();
s.agents_fetched_ms = now_ms();
FN_DBG("fetch thread parsed %zu rows", s.agents.size());
}
}
s.fetching = false;
FN_DBG("fetch thread DONE");
}).detach();
FN_DBG("fetch_agents_async EXIT (thread detached)");
}
// POST action to /agents/{id}/{action}
@@ -484,6 +502,9 @@ static void start_status_sse(AppState& s) {
// ---------------------------------------------------------------------------
static bool g_self_test = false;
static int g_auto_refresh_after_frames = 0; // >0: trigger fetch_agents_async after N frames
static int g_auto_exit_after_frames = 0; // >0: exit after N frames (for headless test)
static int g_frame_count = 0;
static bool run_self_test() {
fn_log::log_info("[self-test] checking subsystems...");
@@ -754,6 +775,20 @@ static void draw_agents_panel(AppState& s) {
};
static const int kSrcForEff[N_COLS] = { 0,1,2,3,4, 5,6,7,8,9,10 };
// First-render init for State: col_visible + col_order sized to N_COLS.
// Without this, render_grid_stage0 indexes into empty std::vector<bool>
// -> undefined behaviour -> Windows access-violation (exit 5).
if ((int)s.agents_tbl_state.col_visible.size() != N_COLS) {
s.agents_tbl_state.col_visible.assign(N_COLS, true);
}
if ((int)s.agents_tbl_state.col_order.size() != N_COLS) {
s.agents_tbl_state.col_order.resize(N_COLS);
for (int c = 0; c < N_COLS; ++c) s.agents_tbl_state.col_order[c] = c;
}
if ((int)s.agents_tbl_state.stages.size() < 1) {
s.agents_tbl_state.stages.resize(1);
}
// Build column specs (Badge for Status, Button for action columns)
static data_table::TableInput main_t;
if (main_t.column_specs.empty()) {
@@ -851,12 +886,27 @@ static void draw_agents_panel(AppState& s) {
// Need at least 1 row for the API to be happy
if (n_rows > 0) {
static int dbg_first = 1;
if (dbg_first) {
FN_DBG("agents_panel PRE-render n_rows=%d cells=%zu specs=%zu eff_h=%p eff_t=%p src=%p vis_sz=%zu",
n_rows, cells_ptr.size(), main_t.column_specs.size(),
(void*)kHeaders, (void*)kTypes, (void*)kSrcForEff, visible_rows.size());
for (int r = 0; r < n_rows && r < 2; ++r) {
for (int c = 0; c < N_COLS; ++c) {
const char* p = cells_ptr[r * N_COLS + c];
FN_DBG(" cell[%d][%d]=%s", r, c, p ? p : "(null)");
}
}
dbg_first = 0;
}
render_grid_stage0("##agents_tbl",
s.agents_tbl_state,
cells_ptr.empty() ? nullptr : cells_ptr.data(),
n_rows, N_COLS, N_COLS,
kHeaders, kTypes, kSrcForEff,
visible_rows, main_t, &events);
static int dbg_post = 1;
if (dbg_post) { FN_DBG("agents_panel POST-render events=%zu", events.size()); dbg_post = 0; }
} else {
ImGui::TextDisabled("(no agents match filter)");
}
@@ -1040,6 +1090,20 @@ static void draw_status_feed_panel(AppState& s) {
// ---------------------------------------------------------------------------
static void render() {
g_frame_count++;
if (g_frame_count <= 3) {
FN_DBG("render frame=%d", g_frame_count);
}
// Headless test: simulate Refresh button click after N frames
if (g_auto_refresh_after_frames > 0 && g_frame_count == g_auto_refresh_after_frames) {
FN_DBG("AUTO-REFRESH triggered at frame %d", g_frame_count);
// Mark connected so the agents panel renders
g_state.connected = true;
// Simulate Connect-then-Refresh: populate base_url default, kick fetch.
fetch_agents_async(g_state);
}
draw_connection_panel(g_state);
if (g_show_agents) draw_agents_panel(g_state);
if (g_show_logs) draw_logs_panel(g_state);
@@ -1052,6 +1116,12 @@ static void render() {
fetch_agents_async(g_state);
}
}
// Headless test: exit after N frames
if (g_auto_exit_after_frames > 0 && g_frame_count >= g_auto_exit_after_frames) {
FN_DBG("AUTO-EXIT at frame %d", g_frame_count);
std::exit(0);
}
}
// ---------------------------------------------------------------------------
@@ -1149,6 +1219,7 @@ static int run_connect_test(const std::string& base_url) {
}
int main(int argc, char** argv) {
FN_DBG("main ENTER argc=%d", argc);
// Self-test mode
for (int i = 1; i < argc; i++) {
if (strcmp(argv[i], "--self-test") == 0) {
@@ -1158,6 +1229,11 @@ int main(int argc, char** argv) {
if (strcmp(argv[i], "--connect-test") == 0 && i + 1 < argc) {
return run_connect_test(argv[i + 1]);
}
if (strcmp(argv[i], "--auto-refresh") == 0) {
g_auto_refresh_after_frames = 30; // ~0.5s @ 60Hz
g_auto_exit_after_frames = 180; // ~3s
FN_DBG("auto-refresh mode enabled");
}
}
if (g_self_test) {
@@ -1186,16 +1262,15 @@ int main(int argc, char** argv) {
// Init DB and load saved base_url + apikey (env first, fallback to `pass agentes/api-key`).
db_open(g_state);
db_load_connection(g_state);
FN_DBG("startup: db loaded base_url=%s", g_state.base_url);
load_apikey(g_state);
if (g_state.apikey_source == "missing") {
fn_log::log_warn("[startup] apikey not found: AGENTS_API_KEY env empty and "
"`pass agentes/api-key` failed. Check GPG agent is unlocked.");
} else {
fn_log::log_info("[startup] apikey loaded from %s", g_state.apikey_source.c_str());
}
FN_DBG("startup: apikey_source=%s apikey_len=%zu",
g_state.apikey_source.c_str(), strlen(g_state.apikey_buf));
// Cleanup on exit
FN_DBG("startup: calling fn::run_app");
int ret = fn::run_app(cfg, render);
FN_DBG("fn::run_app returned %d", ret);
// Persist state
db_save_state(g_state, "log_autoscroll", g_state.log_autoscroll ? "1" : "0");
tests/test_connect_e2e.py
+35
View File
@@ -82,6 +82,41 @@ def test_connect_succeeds_with_valid_apikey():
assert n > 0, f"expected at least 1 agent, got {n}"
def test_app_survives_auto_refresh_cycle():
"""Regression: app must NOT crash on Refresh Agents button click.
Bug history: v0.2 migration to data_table_cpp_viz left State.col_visible
and State.col_order uninitialized — render_grid_stage0 indexed into empty
std::vector<bool>, causing an access violation (Windows exit code 5).
The --auto-refresh CLI flag triggers fetch_agents_async + a full render
cycle from a headless GLFW window, then exits at frame 180 (~3s @ 60Hz).
Exit 0 means the agents panel rendered the live data without crashing.
"""
pass_check = subprocess.run(["pass", "agentes/api-key"],
capture_output=True, text=True, timeout=5)
if pass_check.returncode != 0 or not pass_check.stdout.strip():
pytest.skip("pass agentes/api-key not readable (GPG locked?)")
# WSL → Windows: launch the .exe and let it self-exit after 180 frames.
r = subprocess.run(
[str(_exe()), "--auto-refresh"],
capture_output=True, text=True, timeout=30,
)
assert r.returncode == 0, (
f"app crashed (exit={r.returncode}); last stderr:\n"
+ "\n".join(r.stderr.splitlines()[-20:])
)
# Sanity: stderr must show that fetch_agents reached the parse step.
assert "fetch thread parsed" in r.stderr, (
f"fetch never reached parse; stderr:\n{r.stderr[-1000:]}"
)
# Sanity: render must have completed at least once (POST-render logged).
assert "agents_panel POST-render" in r.stderr, (
f"render_grid_stage0 crashed before completing; stderr:\n{r.stderr[-1000:]}"
)
def test_connect_falls_back_to_pass_when_env_empty():
"""When AGENTS_API_KEY env is empty, the .exe must fetch apikey via
`wsl.exe pass agentes/api-key` (or `pass` on Linux). This is what makes