[PATCH 1/2] Fix endian-dependent test for good
authorStephan Bergmann <sbergman@redhat.com>
Tue, 19 Jul 2022 08:07:57 +0000 (10:07 +0200)
committerRene Engelhard <rene@debian.org>
Wed, 10 Aug 2022 16:04:33 +0000 (17:04 +0100)
commit8c2d3e9787d6e4ae07d5a945c381d2ef469a58fc
treee5df623901cac875d16affbe5658ff1a82be2450
parenta12ea7c39a53991818550c1a7351ba422378161c
[PATCH 1/2] Fix endian-dependent test for good

SvmTest::testComment introduced in e10df88b8ae4138862d4dd25c221189878641aa4 "Add
Comment cppunit test to vcl" calls SvmTest::checkComment twice, first for a
programmatically generated svm document, and then for the canned
vcl/qa/cppunit/svm/data/comment.svm.  4f533655dd635dab30574c2a3b3915377124bd60
"this test is endian-dependant" had fixed the endian-dependent check in
SvmTest::checkComment for the first, programmatically generated case, but broke
it for the second, canned case.  As seen on s390x:

> xmltesttools.cxx:171:Assertion
> Test name: SvmTest::testComment
> equality assertion failed
> - Expected: 00540068006500730065002000610072006500200073006f006d00650020007400650073007400200064006100740061
> - Actual  : 540068006500730065002000610072006500200073006f006d0065002000740065007300740020006400610074006100
> - In <>, attribute 'data' of '/metafile/comment[2]' incorrect value.

So just not make the problematic MetaCommentAction payload endian-dependent in
the first place.  (And one wonders why the test even needs to test such a
problematic payload.)

Change-Id: I203dae6545e0d4e63e7c2d521a27c661e1cf633c
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/137223
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Gbp-Pq: Name svm-test-bigendian.diff
vcl/qa/cppunit/svm/svmtest.cxx