Fixes crashers like the one when cancelling authentication request.
The error message are still not meaningful for the user, but this would
require libcmis API changes.
Change-Id: I22afbf4d39522a2b0dbd043a68dfef2b9308dcec
This also reverts previous attempts at fixing this, commits
33839f90e6 "ucb: try to fix weird STL assertion on
tinderbox" and 6506af86b5 "ucb: second try to fix
weird STL assertion on tinderbox."
Change-Id: I89d0eb87fbd164c0a4cf24d60f225767cb2bfc1a
It's possible that the constructor of RegexpMapIterImpl does not
initialize m_aIndex, so try to park m_aIndex on some end() iterator;
it looks like m_aIndex will not actually be used in this case, because
m_nList = -1.
Change-Id: I74cd0d1d87f3e90217c39bcd3168f157066ff3cb
Hierarchical path is assumed in loads of places, changed the URLs to the
following:
* Binding URL is encoded in the authority part, the repository ID is
set as a fragment of the binding URL.
* The hierarchical path reflects one of the path to the document on
the server
:x
Change-Id: I8214daeb1d9c9b0f6ab86bdf60875e7e4e5369f4
...which happens when installing non-bundle extensions (like a plain .xcu
file), where the path computed for the description.xml contains the path to
the non-bundle extension file itself as a prefix, so access results in
E_NOTDIR.
Change-Id: Ic0f4a34b46d99d1817ceea599bb8b53dbd65bc49
Thist idea caused too intrusive changes to the code of call sites, I
think. Will do it another way that leaves call sites as is.
This reverts commit 25d114eec4.
Start with the first service needed when running sc's filters_test:
the UniversalContentBroker. It might not be typical as ucb uses the
deprecated XSingleServiceFactory.
This commit is not at all a complete solution, far from it, just an
initial hack. Naturally once I come up with some generic enough way to
do the static linking the necessary macros etc will be added to some
general header.
The macro XSERVICEINFO_IMPL_1 (local to ucbhelper and ucb) is amended
so that it in the disable-dynamic-linking case also emits a function
whose assembler name equal contains the service name. This function
returns the XSingleServiceFactory for the service. Where the service
is instantiated we link directly to that function.
But probably this will be reworked a couple of times... Maybe it would
be better to simply have the service name specific entry point be a
pointer to the component's component_getFactory() function? Those all
have the same prototype.
...which has the necessary features to support it.
Change a lot of classes to either contain a protected non-virtual dtor
(which is backwards compatible, so even works for cppumaker-generated
UNO headers) or a public virtual one.
cppuhelper/propertysetmixin.hxx still needs to disable the warning, as
the relevant class has a non-virtual dtor but friends, which would still
cause GCC to warn.
Includes a patch for libcmis, intended to be upstreamed.
