Re: [libvirt] [PATCH 1/2] tests: use virGetLastErrorMessage()
On 05/10/2016 08:16 AM, Jovanka Gulicoska wrote:
---
tests/commandtest.c | 81 ++++++++++++++--------------------------
tests/libvirtdconftest.c | 26 ++++++-------
tests/openvzutilstest.c | 7 +---
tests/securityselinuxlabeltest.c | 6 +--
tests/securityselinuxtest.c | 6 +--
tests/virpolkittest.c | 18 ++++-----
6 files changed, 55 insertions(+), 89 deletions(-)
There's a couple other test instances
- tests/qemucapsprobe.c
- tests/nodeinfo.c:linuxTestCompareFiles : you'll want to do
if (virGetLastError())
VIR_TEST_DEBUG("\n%s\n", error->message);
to preserve the behavior
Comments inline
diff --git a/tests/libvirtdconftest.c b/tests/libvirtdconftest.c
index 61d861d..daa2b90 100644
--- a/tests/libvirtdconftest.c
+++ b/tests/libvirtdconftest.c
@@ -102,7 +102,7 @@ testCorrupt(const void *opaque)
data->params,
data->paramnum,
&type);
- virErrorPtr err = NULL;
+ const char *err = NULL;
if (!newdata)
return -1;
@@ -115,15 +115,15 @@ testCorrupt(const void *opaque)
goto cleanup;
}
- err = virGetLastError();
- if (!err || !err->message) {
+ err = virGetLastErrorMessage();
+ if (!err) {
VIR_DEBUG("No error or message %p", err);
ret = -1;
goto cleanup;
}
virGetLastErrorMessage() doesn't return NULL, so this check will never
trigger. I think it's better to just not change libvirtdconftest.c. Same with
virpolkittest.c which is similar
Thanks,
Cole