Tested on x86_64, i686-darwin and on x86_64-linux where without it whole-1
fails. OK for trunk?
thanks
Iain
--- 8< ---
Whole-1.a68 fails on some test platforms (in some cases reliably, in some cases
sporadically, depending on load). It is close to the 20s limit set by the
algol68
driver.
This adds the capability to use dg-timeout-factor and uses it to double the
allowed
time for this test.
gcc/testsuite/ChangeLog:
* algol68/execute/whole-1.a68: Allow 2x timeout.
* lib/algol68.exp: Add support for dg-timeout-factor.
Signed-off-by: Iain Sandoe <[email protected]>
---
gcc/testsuite/algol68/execute/whole-1.a68 | 2 ++
gcc/testsuite/lib/algol68.exp | 1 +
2 files changed, 3 insertions(+)
diff --git a/gcc/testsuite/algol68/execute/whole-1.a68
b/gcc/testsuite/algol68/execute/whole-1.a68
index ad349776549..8ecb0a01dea 100644
--- a/gcc/testsuite/algol68/execute/whole-1.a68
+++ b/gcc/testsuite/algol68/execute/whole-1.a68
@@ -1,3 +1,5 @@
+{ This test takes more than 20s on several platforms. }
+{ dg-timeout-factor 2 }
begin
{ RR standard proc whole replaced; test it }
diff --git a/gcc/testsuite/lib/algol68.exp b/gcc/testsuite/lib/algol68.exp
index 9953fe27794..e9021a09626 100644
--- a/gcc/testsuite/lib/algol68.exp
+++ b/gcc/testsuite/lib/algol68.exp
@@ -21,6 +21,7 @@
load_lib prune.exp
load_lib gcc-defs.exp
load_lib timeout.exp
+load_lib timeout-dg.exp
load_lib target-libpath.exp
#
--
2.50.1 (Apple Git-155)