This is nice to make them more easily debuggable.
A series of crude hacks are employed to bootstrap enough services from
python so the current tests run.
This is only tested with system python3 on Fedora.
Change-Id: I5e06741e55ead7fddec41ff776ff8ca5d2399469
Reviewed-on: https://gerrit.libreoffice.org/3215
Reviewed-by: David Ostrovsky <David.Ostrovsky@gmx.de>
Tested-by: David Ostrovsky <David.Ostrovsky@gmx.de>
...which effectively is just a glorious wrapper around
comphelper::getProcessServiceFactory.
In turn gets also rid of ImplSVAppData's mxMSF and mpMSFTempFileName and the
rSMgr parameter to InitVCL.
All the VCL users "soffice", "spadmin", and "unopkg gui" appear to still work
fine.
Change-Id: I797d48f7d0d8c35bb82124c9ab0ee63850c4d863
* As UCB is only ever initialized with "Local"/"Office", remove this
configuration vector completely. The "create" ctor creates an instance
internally initialized with those "Local"/"Office" keys. Special (test) code
can still instantiate an uninitialized one via plain createInstance. And for
backwards compatilibity process startup still ensures to create an initialized
instance early, in case there is still code out there (in extensions) that
later calls plain createInstance and expects to get the already-initialized
(single) instance.
* XInitialization is an "implementation detail" of the UniversalContentBroker
service, do not expose in XUniversalContentBroker.
* ucbhelper/configurationkeys.hxx is no longer needed and is removed.
* ucbhelper/contentbroker.hxx is an empty wrapper and is removed; however, that
requires ucbhelper::Content constructors to take explicit XComponentContext
arguments now.
* The only remaining code in ucbhelper/source/client/contentbroker.cxx is
Android-only InitUCBHelper. Is that relevant still?
Change-Id: I3f7bddd0456bffbcd13590c66d9011915c760f28
avoids the problems of dangling uno singletons invalidated after the first
dispose and the chain of other singletons that don't expect to need to
re-initialize, etc.
reenable editeng cppunit test
inherit i18npool cppunit test from unotest base
drop LibreOfficeProtector, do "throwable" work in setUp/tearDown not
in ctors/dtors
sot is below vcl, but is a filters test, we can split bootstrapfixture
into a vcl needing bit and and non-vcl bit and filters test api
can be standalone and combined with whichever bit in order to form
pre and post vcl filters test
It turns out the constructors are all run back-to-back one per filter,
and we need to init and de-init in a sensible order to make multiple
tests that otherwise share the ContentBroker work properly.
