diff options
author | Stephan Bergmann <sbergman@redhat.com> | 2020-03-11 16:06:05 +0100 |
---|---|---|
committer | Stephan Bergmann <sbergman@redhat.com> | 2020-03-16 10:14:35 +0100 |
commit | 6ddec950dc642dcf07956cce42c450a0b775963f (patch) | |
tree | f7d2831bf2429b3301ab2a0e59c115440c0faba4 /include/xmloff/XMLFontAutoStylePool.hxx | |
parent | cc17485472ca4217013cebaa32f7b574c020a394 (diff) |
Avoid -Werror=maybe-uninitialized
...as seen at least with GCC 10 trunk and --enable-optimized, when bMirroredX/Y
were used without checking that reading them with >>= had succeeded.
Assuming that in this code (added with b4a6977e05d87fe0a79b266ec30e4f403404f1b4
"tdf#129532 tdf#98839 fixes for mirror of custom shapes") it cannot be
guaranteed that rItem.Value is of the right type (so o3tl::forceAccess, which
uses assert to verify the right type, would not be appropriate), but that it is
an exception-worthy error if rItem.Value is not of the right type (so
o3tl::doAccess, which throws a css::uno::RuntimeException upon a wrong type, is
appropriate).
Change-Id: Ibe6991d69a1d6a0c2c41c839c240050a6355e98b
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/90337
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Diffstat (limited to 'include/xmloff/XMLFontAutoStylePool.hxx')
0 files changed, 0 insertions, 0 deletions