Source: cppformat
Version: 1.1.0+ds-3
Severity: normal
Tags: patch
User: debian-powe...@lists.debian.org
Dear Maintainer,
Although this package compiles on ppc64el, it fails in the test suite.
Specifically, the format-test fails in FormatterTest.FormatLongDouble.
This is what is in the log:
Value of: format("{0:+010.4g}", 392.65l)
Actual: "+0000392.7"
Expected: "+0000392.6"
>From printing the number as a float, with format specifier %.20LF, I could
see that 392.65l in ppc64el is rounded from 392.65000000000000000000, while
in x86_64 it is rounded from 392.64999999999999999445.
I can't really explain the reason for the difference, so the patch I am
attaching here just checks the architecture in precompilation and the test
then expects the number rounded accordingly.
Thanks & Regards
Fernando
-- System Information:
Debian Release: 8.0
APT prefers stable-updates
APT policy: (500, 'stable-updates'), (500, 'unstable'), (500, 'stable'), (1,
'experimental')
Architecture: ppc64el (ppc64le)
Kernel: Linux 3.16.0-4-powerpc64le (SMP w/8 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)
--- cppformat-1.1.0+ds.orig/test/format-test.cc
+++ cppformat-1.1.0+ds/test/format-test.cc
@@ -1163,7 +1163,11 @@ TEST(FormatterTest, FormatLongDouble) {
safe_sprintf(buffer, "%Le", 392.65l);
EXPECT_EQ(buffer, format("{0:e}", 392.65l));
safe_sprintf(buffer, "%LE", 392.65l);
+#ifdef __PPC__
+ EXPECT_EQ("+0000392.7", format("{0:+010.4g}", 392.65l));
+#else
EXPECT_EQ("+0000392.6", format("{0:+010.4g}", 392.65l));
+#endif
}
TEST(FormatterTest, FormatChar) {
