@wendellresearch-ai/benchwarmers 0.5.4 → 0.5.5

package/VERSION

@@ -1 +1 @@
-0.5.4
+0.5.5

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@wendellresearch-ai/benchwarmers",
-  "version": "0.5.4",
+  "version": "0.5.5",
   "description": "Self-hosted panel gateway for LLMs: ask a collection of models at once, keep every answer side by side, rotate explorer models in, and let your verdicts pick the next panel.",
   "mcpName": "io.github.WendellResearch-AI/benchwarmers",
   "license": "MIT",

package/scripts/test_serve_edit.py

@@ -461,6 +461,25 @@ def main() -> int:
     print(("ok   " if ok else "FAIL ") + "action needed drops alerts on retired models")
     failures += 0 if ok else 1
 
+    # Retry is one request: no client retries after a 429 (2026-11-02: each
+    # UnoRouter Retry waited its 60 s and sent a second, refused, request).
+    import run
+    asked, kept_ask = {}, run.ask_once
+
+    def fake_ask(*a, **kw):
+        asked.update(kw)
+        raise StopIteration                      # stop before anything is written
+    run.ask_once = fake_ask
+    try:
+        serve.retry_model(next(iter(serve.registered().values())))
+    except StopIteration:
+        pass
+    finally:
+        run.ask_once = kept_ask
+    ok = asked.get("retries") == 0
+    print(("ok   " if ok else "FAIL ") + "Retry sends one request, no client retries" + ("" if ok else f" -> {asked}"))
+    failures += 0 if ok else 1
+
     # A provider's words end where its link begins, and are not cut short here:
     # the page cuts them to a line, and a phone, with no hover, shows them whole.
     text, url = serve.said("Error: Error code: 403 - {'error': {'message': 'm needs your consent. Give it once at"

package/serve.py

@@ -701,12 +701,17 @@ def retry_model(alias: str, who: str = "ui") -> dict:
     """One fresh call to a model on a retiring alert. An answer is written to
     perf/history.csv as evidence (scripts/auto_retire.py reads it, so the
     clock resets) and the alert is cleared now rather than at the next 5am
-    check. A failure writes nothing: the daily probe keeps the record."""
+    check. A failure writes nothing: the daily probe keeps the record.
+
+    One request, never the client's retries: a 429's stated wait (UnoRouter's
+    60 s) then a second request spent a minute and another call of a 1-a-minute
+    free tier, and Retry all did that for each model in turn. A person can
+    click again."""
     names = registered()
     if alias not in names.values():
         return {"ok": False, "errors": [f"{alias!r} is not registered"]}
     from run import ask_once, failure_class, llm_binary
-    body, seconds = ask_once(llm_binary(), alias, "Reply with exactly one word: ready", 90)
+    body, seconds = ask_once(llm_binary(), alias, "Reply with exactly one word: ready", 90, retries=0)
     cls = failure_class(body, min_length=1)
     if cls is not None:
         try:                                           # a stated quota or wait still teaches a cooldown
@@ -2129,21 +2134,31 @@ def actions_page(groups: dict, auto: list) -> str:
         shown = re.sub(r"^https?://", "", url).split("?")[0]
         return f" <a href='{esc(url)}' target='_blank' rel='noopener'>{esc(shown)} ↗</a>"
 
+    def why(text: str, url: str, title: str) -> str:
+        return (f"<td class='why' title='{esc(title)}'><div class='said-line'><span class='said'>{esc(text)}</span>"
+                f"{link(url)}</div></td>")
+
     def model_row(r: dict) -> str:
         if not r["alias"]:                             # a notice: what it says, and Got it if it goes once seen
             got = (f"<button class='linkish' data-act='dismiss' data-key='{esc(r['key'])}'>Got it</button>"
                    if r.get("dismiss") else "")
-            return (f"<div class='a-row' data-row><span class='what'>{esc(r['said'])}</span>"
-                    f"<span>{chip(r['label'], r['css'])}</span><span class='why'>{esc(r['title'])}</span>"
-                    f"<span class='acts'>{got}</span></div>")
+            return (f"<tr data-row><td class='what'>{esc(r['said'])}</td><td>{chip(r['label'], r['css'])}</td>"
+                    f"<td class='why'>{esc(r['title'])}</td><td class='acts'>{got}</td></tr>")
         alias = esc(r["alias"])
-        return (f"<div class='a-row' data-row><a href='#' class='model' data-alias='{alias}'>"
-                f"<span class='alias'>{breakable(alias)}</span></a><span>{chip(r['label'], r['css'])}</span>"
-                f"<span class='why' title='{esc(r['title'])}'><span class='said'>{esc(r['said'])}</span>{link(r['url'])}</span>"
-                f"<span class='acts'><button class='linkish' data-act='retry' data-alias='{alias}'"
+        return (f"<tr data-row><td><a href='#' class='model' data-alias='{alias}'><span class='alias'>{breakable(alias)}"
+                f"</span></a></td><td data-sort='{r['days']:03d}{esc(r['label'])}'>{chip(r['label'], r['css'])}</td>"
+                + why(r["said"], r["url"], r["title"])
+                + f"<td class='acts'><button class='linkish' data-act='retry' data-alias='{alias}'"
                 " title='One call now; an answer clears the alert'>Retry</button>"
                 f"<button class='linkish danger' data-act='retire' data-alias='{alias}'"
-                f" data-why='{esc(r['why'] or r['label'])}'>Retire</button></span></div>")
+                f" data-why='{esc(r['why'] or r['label'])}'>Retire</button></td></tr>")
+
+    # One column plan for every table, as on Models, so they line up down the page.
+    cols = "<colgroup><col class='c-model'><col class='c-state'><col><col class='c-acts'></colgroup>"
+
+    def table(first: str, second: str, rows: str) -> str:
+        return (f"<table class='a-table'>{cols}<thead><tr>{th(first)}{th(second)}{th('Reason')}<th></th></tr></thead>"
+                f"<tbody>{rows}</tbody></table>")
 
     def section(conn: str, g: dict) -> str:
         aliases = [r["alias"] for r in g["rows"] if r["alias"]]
@@ -2156,29 +2171,30 @@ def actions_page(groups: dict, auto: list) -> str:
             f"<div class='a-held' title='{esc(h['title'])}'><span class='chip'>held</span><span>Every model"
             f" here is failing, likely the connection (key, credit, prefix), so nothing was retired."
             f" Latest: {esc(h['said'])}{link(h['url'])}</span></div>" for h in g["held"])
-        n = len(g["rows"])
-        return (f"<div class='a-group'><div class='a-head' data-row><span class='conn'>{esc(conn)}</span>"
-                f"<span class='count'>{n} {'notice' if conn == 'Notices' else 'model'}{'s' * (n != 1)}</span>"
-                f"<span class='why'></span><span class='acts'>{both}</span></div>{held}"
-                f"<div class='a-rows'>{''.join(model_row(r) for r in g['rows'])}</div></div>")
+        n, notices = len(g["rows"]), conn == "Notices"
+        return (f"<div class='panel'><div class='a-head' data-row><h3 class='chart-head'>{esc(conn)} · {n}"
+                f" {'notice' if notices else 'model'}{'s' * (n != 1)}</h3><span class='why'></span>"
+                f"<span class='acts'>{both}</span></div>{held}"
+                + table("Notice" if notices else "Model", "Kind" if notices else "State",
+                        "".join(model_row(r) for r in g["rows"])) + "</div>")
 
     tags = load_json(DATA / "config" / "provider-tags.json")      # a retired model's connection, by served name
     gone = "".join(
-        f"<div class='a-row' data-row><a href='#' class='model' data-alias='{esc(alias)}'>"
-        f"<span class='alias'>{breakable(esc(alias))}</span></a><span class='when'>{esc(r['at'][:10])}</span>"
-        f"<span class='why' title='{esc(r.get('why', ''))}'><span class='said'>"
-        f"{esc(tags.get(r.get('served', ''), '').strip('?') or r.get('endpoint') or endpoints.default())}"
-        f" · {esc(said(r.get('why', '').removeprefix('auto: '))[0])}</span></span>"
-        f"<span class='acts'><button class='linkish' data-act='restore' data-alias='{esc(alias)}'>Restore</button></span></div>"
+        f"<tr data-row><td><a href='#' class='model' data-alias='{esc(alias)}'>"
+        f"<span class='alias'>{breakable(esc(alias))}</span></a></td><td class='when'>{esc(r['at'][:10])}</td>"
+        + why(f"{tags.get(r.get('served', ''), '').strip('?') or r.get('endpoint') or endpoints.default()}"
+              f" · {said(r.get('why', '').removeprefix('auto: '))[0]}", "", r.get("why", ""))
+        + f"<td class='acts'><button class='linkish' data-act='restore' data-alias='{esc(alias)}'>Restore</button></td></tr>"
         for alias, r in auto)
     # Folded: a week of automatic retirements is a record to check, not a list of things to do.
-    folded = (f"<details class='a-retired'><summary>Retired automatically this week · {len(auto)}"
-              f" <span class='muted'>— each can be restored</span></summary><div class='a-rows'>{gone}</div></details>"
-              if auto else "")
-    body = "".join(section(conn, g) for conn, g in groups.items()) or "<p class='a-none'>Nothing open.</p>"
+    folded = (f"<div class='panel'><details class='a-retired'><summary><h3 class='chart-head'>Retired automatically"
+              f" this week · {len(auto)}</h3> <span class='muted'>each can be restored</span></summary>"
+              f"{table('Model', 'Retired', gone)}</details></div>" if auto else "")
+    body = "".join(section(conn, g) for conn, g in groups.items())
     return ("<div class='panel'><h2>Action needed</h2><p class='sub'>States that will not fix themselves, by the"
             " connection they come from. A reason's full error is on hover, and a name opens the model."
-            f" Retry spends one call; an answer clears the alert.</p>{body}{folded}</div>")
+            " Retry spends one call; an answer clears the alert.</p>"
+            + ("" if body else "<p class='a-none'>Nothing open.</p>") + f"</div>{body}{folded}")
 
 
 CAP_LABELS = {"tools": "tools", "vision": "vision", "structured_output": "json", "reasoning": "reasons"}

package/suite/run.py

@@ -920,9 +920,11 @@ def _child_env() -> dict:
 
 
 def ask_once(binary: str, model: str, prompt: str,
-             timeout: int, images: list[str] | None = None) -> tuple[str, float]:
+             timeout: int, images: list[str] | None = None, retries: int | None = None) -> tuple[str, float]:
     # BENCH_CLIENT=builtin: the same call without the `llm` subprocess, giving
-    # the same text for the same failure (metrics/client.py).
+    # the same text for the same failure (metrics/client.py). `retries`: the
+    # built-in client's tries after a 429, 5xx or connection error (its two
+    # by default); the llm backend keeps its SDK's own.
     if str(ROOT / "metrics") not in sys.path:
         sys.path.insert(0, str(ROOT / "metrics"))
     import client
@@ -935,7 +937,7 @@ def ask_once(binary: str, model: str, prompt: str,
                           purpose="vision task")
         return client.output(res, timeout), res["seconds"]
     if client.backend(model) == "builtin":
-        res = client.chat(model, prompt, timeout=timeout)
+        res = client.chat(model, prompt, timeout=timeout, retries=retries)
         return client.output(res, timeout), res["seconds"]
     started = time.monotonic()
     try:

package/ui/bench.css

@@ -115,16 +115,11 @@ main { max-width: var(--page, 104rem); margin: 0 auto; padding: 0 var(--gutter)
 .problems .head { display: flex; align-items: baseline; gap: .5rem; }
 .problems .head h2 { margin-bottom: .25rem; }
 
-/* Action needed: its own page, by connection. Each model is one line of a
-   grid shared by its group (name, state, the provider's words, actions), so
-   the columns line up down the group. */
+/* Action needed: its own page, a panel per connection as on Models, each a
+   table (name, state, the provider's words, actions). */
 .a-line { display: flex; flex-wrap: wrap; align-items: baseline; gap: .25rem .6rem; margin: 1.4rem 1.6rem 0;
           padding: .45rem .75rem; background: var(--err-ground); border-radius: var(--radius); }
-.a-group { padding: 1rem 0 .75rem; border-top: 1px solid var(--line); }
-.a-group:first-of-type { border-top: 0; padding-top: .5rem; }
-.a-head { display: flex; flex-wrap: wrap; align-items: baseline; gap: .25rem .6rem; margin-bottom: .4rem; }
-.a-head .conn { font-size: 1.0625rem; font-weight: 700; }
-.a-head .count { color: var(--muted); }
+.a-head { display: flex; flex-wrap: wrap; align-items: baseline; gap: .25rem .6rem; }
 .a-head .why { font-size: .875rem; }
 .a-head .why:empty { display: none; }
 .a-head .acts { margin-left: auto; display: flex; gap: .5rem; }
@@ -132,26 +127,26 @@ main { max-width: var(--page, 104rem); margin: 0 auto; padding: 0 var(--gutter)
           background: var(--warn-ground); border-radius: var(--radius); font-size: .9375rem; overflow-wrap: anywhere; }
 .a-held .chip { flex: none; background: var(--ink); color: var(--paper); }
 .a-held a { color: var(--ink); font-weight: 500; }
-/* The same fixed columns in every group and the retired list, so state,
-   words and actions line up down the whole page whatever each group's
-   longest name; a long name wraps in its column. */
+/* The same fixed columns in every table, so state, words and actions line
+   up down the whole page whatever each group's longest name; a long name
+   wraps in its column. */
 #tab-needs > .panel { container-type: inline-size; }
-.a-rows { display: grid; grid-template-columns: minmax(0, 32%) 8.25rem minmax(0, 1fr) 6.25rem;
-          gap: 0 1rem; align-items: baseline; }
-.a-row { display: contents; }
-.a-row > * { padding: .3rem 0; border-top: 1px solid color-mix(in oklab, var(--line) 55%, transparent); }
-.a-row:first-child > * { border-top: 0; }
-.a-row .alias, .a-row .what { font-weight: 500; overflow-wrap: anywhere; }   /* a name breaks at its dots first (<wbr>) */
-.a-row .why, .a-row .when { color: var(--muted); font-size: .875rem; overflow-wrap: anywhere; }
+table.a-table { table-layout: fixed; }
+table.a-table col.c-model { width: 30%; }
+table.a-table col.c-state { width: 9.5rem; }
+table.a-table col.c-acts { width: 7.5rem; }
+.a-table .alias, .a-table .what { font-weight: 500; overflow-wrap: anywhere; }   /* a name breaks at its dots first (<wbr>) */
+.a-table .why, .a-table .when { color: var(--muted); font-size: .875rem; overflow-wrap: anywhere; }
 /* One line a model: the provider's words cut to the column, whole on hover; the link gives way last. */
-.a-row .why:has(> .said) { display: flex; gap: .5rem; min-width: 0; }
-.a-row .said { flex: 0 1 auto; min-width: 0; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
-.a-row .why > a { flex: none; max-width: 50%; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
-.a-row .acts { display: flex; gap: .9rem; font-size: .875rem; }
+.a-table .said-line { display: flex; gap: .5rem; min-width: 0; }
+.a-table .said { flex: 0 1 auto; min-width: 0; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
+.a-table .said-line > a { flex: none; max-width: 50%; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
+.a-table td.acts { white-space: nowrap; font-size: .875rem; }
+.a-table td.acts .linkish + .linkish { margin-left: .9rem; }
 .linkish.danger { color: var(--err); }
-.a-retired { margin-top: .75rem; border-top: 1px solid var(--line); padding-top: .5rem; }
-.a-retired > summary { cursor: pointer; font-weight: 500; padding: .35rem 0; }
-.a-retired > summary .muted { color: var(--muted); font-weight: 400; }
+.a-retired > summary { cursor: pointer; }
+.a-retired > summary h3 { display: inline; }
+.a-retired > summary .muted { color: var(--muted); }
 .a-none { color: var(--muted); }
 
 /* pages: a grouped sidebar (Observe, Gateway, Control); the group names and
@@ -550,13 +545,14 @@ dialog .pick { display: flex; gap: 1.5rem; }
    state on one line, the words and actions under them. No gap between name
    and state, so the line above a model runs unbroken. */
 @container (max-width: 46rem) {
-  .a-rows { grid-template-columns: minmax(0, 1fr) max-content; column-gap: 0; }
-  .a-row > :nth-child(2) { padding-left: .75rem; }
-  .a-row .why, .a-row .acts { grid-column: 1 / -1; padding-top: 0; border-top: 0; }
-  .a-row .acts { padding-bottom: .5rem; }
-  .a-row .why:has(> .said) { flex-wrap: wrap; }
-  .a-row .said { white-space: normal; }             /* the height to read it whole */
-  .a-row .why > a { max-width: none; }
+  .a-table thead { display: none; }
+  .a-table, .a-table tbody { display: block; }
+  .a-table tr { display: grid; grid-template-columns: minmax(0, 1fr) max-content; border-bottom: 1px solid var(--line); }
+  .a-table td { border-bottom: 0; }
+  .a-table td.why, .a-table td.acts { grid-column: 1 / -1; padding-top: 0; }
+  .a-table .said-line { flex-wrap: wrap; }
+  .a-table .said { white-space: normal; }           /* the height to read it whole */
+  .a-table .said-line > a { max-width: none; }
 }
 @media (prefers-reduced-motion: reduce) {
   * { transition: none !important; }