- edited description
tests/bug33.phpt is slow on s390x
I’m using to package the extension 1.1.0
(PHP8) for Alpinelinux but running tests on s390x architecture I’m getting very slow execution of bug33
test
using patch to debug numbers
--- a/tests/bug33.phpt
+++ b/tests/bug33.phpt
@@ -32,6 +32,9 @@
if ($time_spent >= $time_estimated && ($time_spent - $time_estimated) <= $max_time_diff) {
echo "ok\n";
}
+else {
+echo "spend $time_spent estimated $time_estimated max $max_time_diff \n";
+}
?>
--EXPECT--
1
I’m getting in CI
TEST 18/19 [tests/bug33.phpt]
========DIFF========
--
8
9
10
011+ spend 0.51614785194397 estimated 0.5 max 0.008
011- ok
========DONE========
FAIL Check for bug #33 [tests/bug33.phpt]
Comments (8)
-
reporter -
reporter Build on mips64 showed other test fail
TEST 10/19 [tests/10_signal.phpt] │ ========DIFF======== │ -- │ ok 12 │ ok 13 │ ok 14 │ 015+ int(16) │ 015- int(10) │ ========DONE======== │ FAIL Check for EvSignal functionality [tests/10_signal.phpt]
No idea how to fix yet
-
reporter -
repo owner @Andy Postnikov ,
Build on mips64 showed other test fail
015+ int(16) │ 015- int(10)
Well, that looks like SIGUSR1 has a different value there (16). So 10 needs to be replaced with a pattern, and EXPECT – with EXPECTF. I’ll do that.
-
repo owner I’m using to package the extension
1.1.0
(PHP8) for Alpinelinux but running tests on s390x architecture I’m getting very slow execution ofbug33
test011+ spend 0.51614785194397 estimated 0.5 max 0.008
The difference of 0.016 doesn't seem critical to me, so I think that increasing the maximum won't break anything. I guess a maximum of 0.1 will be enough.
-
repo owner - changed status to resolved
Fixed
#39: tests/bug33.phpt is slow on s390x→ <<cset 94d83224e12c>>
-
repo owner Should be fixed in version 1.1.1-beta. Let's see.
-
reporter Thank you a lot! Everything builds and passes check now
- Log in to comment