You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
gap suite red on main: test_gap_fetch_request_from_node_incoming_message aborts with 'there is no reactor running' (perry-ext-http/server.rs:911) #7932
Found while running the gap suite for #7916 (PR #7928). Not caused by that branch — it
reproduces on a clean origin/main build, deterministically, 3/3 runs.
$ ./test_gap_fetch_request_from_node_incoming_message
thread '<unnamed>' panicked at crates/perry-ext-http/src/server/server.rs:911:13:
there is no reactor running, must be called from the context of a Tokio 1.x runtime
Abort trap: 6 # exit 134
Perry prints zero lines before aborting; node prints
len=55 match=true
expected len=55
and exits 0.
test_gap_fetch_request_from_node_incoming_message is not in test-parity/gap_snapshot.json, so the gate expects it to pass and scripts/run_gap_tests.sh is red on main because of it.
Provenance (A/B'd, not assumed)
Two compilers, one from origin/main @ ebefba51a and one from the #7928 branch, three
runs each: exit=134 in all six. Same panic message.
Parity is tag-gated (it has not run per-PR since v0.5.1018), which is presumably how this
landed and stayed. Sibling finding, different root cause: #7930.
It is six tests, not one, and it is the whole HTTP/net server family. Full gap-suite run
on origin/main @ ebefba51a, every one A/B'd against a second compiler with
byte-identical stdout and identical exit codes, so none of them belongs to the branch that
found them:
test
exit
panic site
test_gap_fetch_request_from_node_incoming_message
134
perry-ext-http/src/server/server.rs:911
test_gap_http_client_no_redirect_follow
134
perry-ext-http/src/server/server.rs:911
test_gap_http_overloads_3226plus
134
perry-ext-http/src/server/server.rs:911
test_gap_http_req_async_iterator
134
perry-ext-http/src/server/server.rs:911
test_gap_http_res_socket_writable_onfinished
134
perry-ext-http/src/server/server.rs:911
test_gap_net_connect_bound_value
134
tokio-1.53.1/src/net/tcp/listener.rs:304
Five share the exact panic site; net_connect_bound_value hits the same "no reactor
running" assertion one frame lower, inside tokio's own TcpListener. All six print zero
lines before aborting.
None of the six is in test-parity/gap_snapshot.json, so all six read as pass -> crash regressions against the committed snapshot.
Closing as a duplicate of #7629. This is the same release-profile ext-staticlib link defect already reproduced and root-caused there: the shared HTTP test aborts at the same TcpListener::from_std reactor assertion, and the five additional HTTP/net failures are the same server-family manifestation. I am taking #7629 for the actual linker fix and will cover all six tests there.
Activity
proggeramlug commentedon Aug 12, 2026
It is six tests, not one, and it is the whole HTTP/net server family. Full gap-suite run
on
origin/main@ebefba51a, every one A/B'd against a second compiler withbyte-identical stdout and identical exit codes, so none of them belongs to the branch that
found them:
test_gap_fetch_request_from_node_incoming_messageperry-ext-http/src/server/server.rs:911test_gap_http_client_no_redirect_followperry-ext-http/src/server/server.rs:911test_gap_http_overloads_3226plusperry-ext-http/src/server/server.rs:911test_gap_http_req_async_iteratorperry-ext-http/src/server/server.rs:911test_gap_http_res_socket_writable_onfinishedperry-ext-http/src/server/server.rs:911test_gap_net_connect_bound_valuetokio-1.53.1/src/net/tcp/listener.rs:304Five share the exact panic site;
net_connect_bound_valuehits the same "no reactorrunning" assertion one frame lower, inside tokio's own
TcpListener. All six print zerolines before aborting.
None of the six is in
test-parity/gap_snapshot.json, so all six read aspass -> crashregressions against the committed snapshot.proggeramlug commentedon Aug 12, 2026
Closing as a duplicate of #7629. This is the same release-profile ext-staticlib link defect already reproduced and root-caused there: the shared HTTP test aborts at the same
TcpListener::from_stdreactor assertion, and the five additional HTTP/net failures are the same server-family manifestation. I am taking #7629 for the actual linker fix and will cover all six tests there.docs: gap-suite triage — six crashes signature-matched to #7932, two …
feat(runtime): #6759 C3 rung 1 — make the shape word uniform (#7983)