osascript enumeration: raise timeout to 30s and record blind spots

29475ed6ef9dc1630f77d72c079873d18d61170f · 2026-09-10 08:33:42 -0700 · Steve Abrams

The iTerm session enumeration ran osascript with timeout=8 and swallowed
failure as `return {}, "unavailable"` -- no exception, no exit code. That is
the false-green class: on a box that hit load 53.78 today (with the sibling
ps call measured at 20.4s), AppleScript enumeration cost scales with tab
count (~49 here), so an 8s ceiling was trippable and its failure was
indistinguishable from "no sessions".

It matters because `start` treats "unavailable" as an empty result and falls
through to painting green instead of restoring a sticky semantic dot -- a
confidently-wrong paint, worse than a missing one.

Raised to 30s, deliberately NOT the 60s used for the ps call: this runs at
session start, where a longer hang would delay startup.

Blind spots now append to the terminal-status-health log via a fail-safe
recorder that never raises and never blocks painting.

DTD 2026-09-10 verdict C (4/5). Reversible: git revert.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PXNMS1TvMiVbE3ckhSaLeT

Files touched

M terminal_status.py

Diff

commit 29475ed6ef9dc1630f77d72c079873d18d61170f
Author: Steve Abrams <steve@designerwallcoverings.com>
Date:   Thu Sep 10 08:33:42 2026 -0700

    osascript enumeration: raise timeout to 30s and record blind spots
    
    The iTerm session enumeration ran osascript with timeout=8 and swallowed
    failure as `return {}, "unavailable"` -- no exception, no exit code. That is
    the false-green class: on a box that hit load 53.78 today (with the sibling
    ps call measured at 20.4s), AppleScript enumeration cost scales with tab
    count (~49 here), so an 8s ceiling was trippable and its failure was
    indistinguishable from "no sessions".
    
    It matters because `start` treats "unavailable" as an empty result and falls
    through to painting green instead of restoring a sticky semantic dot -- a
    confidently-wrong paint, worse than a missing one.
    
    Raised to 30s, deliberately NOT the 60s used for the ps call: this runs at
    session start, where a longer hang would delay startup.
    
    Blind spots now append to the terminal-status-health log via a fail-safe
    recorder that never raises and never blocks painting.
    
    DTD 2026-09-10 verdict C (4/5). Reversible: git revert.
    
    Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
    Claude-Session: https://claude.ai/code/session_01PXNMS1TvMiVbE3ckhSaLeT
---
 terminal_status.py | 47 +++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 45 insertions(+), 2 deletions(-)

diff --git a/terminal_status.py b/terminal_status.py
index 03068f0..b7e4613 100644
--- a/terminal_status.py
+++ b/terminal_status.py
@@ -31,6 +31,39 @@ COLORS = {
 PRIORITY = {name: i for i, name in enumerate(
     ("orange", "purple", "yellow", "green", "pink", "none"))}
 TTY = re.compile(r"ttys[0-9]+")
+
+
+def _record_enum_blind(reason):
+    """Record that iTerm session enumeration went blind (TK-11369).
+
+    A timeout here used to return ({}, "unavailable") silently, which makes
+    `start` paint green instead of restoring a sticky semantic dot. Never
+    raises and never blocks: a failure to log must not break painting.
+    """
+    try:
+        d = os.path.join(os.path.expanduser("~"),
+                         ".claude/skills/terminal-status-health/data")
+        os.makedirs(d, exist_ok=True)
+        line = json.dumps({
+            "ts": time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()),
+            "pid": os.getpid(),
+            "rc": "enum_blind",
+            "err": str(reason)[:300],
+        }) + "\n"
+        path = os.path.join(d, "repaint-failures.jsonl")
+        try:
+            if os.path.getsize(path) > 262144:
+                os.replace(path, path + ".1")
+        except OSError:
+            pass
+        # Single O_APPEND write keeps the record atomic across concurrent writers.
+        fd = os.open(path, os.O_WRONLY | os.O_CREAT | os.O_APPEND, 0o644)
+        try:
+            os.write(fd, line.encode())
+        finally:
+            os.close(fd)
+    except Exception:
+        pass
 VERSION = 1
 
 
@@ -466,11 +499,21 @@ tell application "iTerm2"
 end tell
 """
     try:
+        # TK-11369 (DTD 2026-09-10, verdict C): was timeout=8. iTerm AppleScript
+        # enumeration cost scales with tab count (~49 here) and this box has hit
+        # load 53.78, so an 8s ceiling was trippable. Raised to 30s -- NOT 60s
+        # like the ps call, because this runs at session start and a longer hang
+        # would delay startup. A timeout here is a genuine blind spot, not a
+        # benign empty result: it makes `start` fall through and paint green
+        # instead of restoring a sticky semantic dot (a confidently-wrong paint),
+        # so it is now recorded rather than silently swallowed.
         result = subprocess.run(["osascript", "-e", script], capture_output=True,
-                                text=True, timeout=8)
+                                text=True, timeout=30)
         if result.returncode:
+            _record_enum_blind("osascript returncode " + str(result.returncode))
             return {}, "unavailable"
-    except (OSError, subprocess.TimeoutExpired):
+    except (OSError, subprocess.TimeoutExpired) as exc:
+        _record_enum_blind(type(exc).__name__ + ": " + str(exc)[:120])
         return {}, "unavailable"
     sessions = {}
     for line in result.stdout.splitlines():

← e6a426b set: automatic paints yield to sticky semantic dots (orange/ · back to Terminal Status · ticket_binding: a slow ps must degrade the label, not kill t fea7539 →