... to clean out all sorts of inconsitencies that can have accrued there over
time apparently due to how the mechanism of copying share/prereg/bundled worked,
and to work around stale $BUNDLED_EXTENSIONS_PREREG references in a better way
than the previous 81fd6b084b "fdo#53006 Force
reinstall of all bundled extensions on upgrade."
See the comment on refreshBundledExtensionsDir for how, at least in theory, that
functionality could be removed again in the future.
This effectively reverts 2d2b19dea1 "Force
ExtensionManager resync when the implementation changes" and
81fd6b084b "fdo#53006 Force reinstall of all
bundled extensions on upgrade" (the latter at least on master; it had never been
cherry-picked to libreoffice-3-6), which it obsoletes.
Change-Id: I8f80c07a06ec9d53b03813338eeff7d7757c9d4d
... in soffice_main calling GetCommandLineArgs, before Desktop::Init takes care
of synchronization of the per-user bundled/shared extension information (as the
latter potentially modifies data that the service manager consumes upon
instantiation; this e.g. lead to bundled extensions not working during first
start after an upgrade).
The only reason GetCommandLineArgs needed ensureProcessServiceFactory appears to
be the ExternalUriReferenceTranslator. So defer its usage to when the relevant
cmd line args are actually processed (which, luckily, does not yet happen in
soffice_main's usage of GetCommandLineArgs).
Change-Id: I6ebbf0a4ad1c6f64c8fbbe2b0d7628fa42a1afb6
Due to the scourge of symlinked git repositories, building the
configuration files in dictionaries fails on cygwin because the native
xsltproc cannot understand cygwin symlinks; work around that with a
new gb_Helper_symlinked_native function.
Should be reverted once git sub modules liberate us from the scourge.
Change-Id: Idc48da9b60348dc66160219909479c3c83daeaf2
a) fix vmldrawing.vml for xlsx export ( changed from frame to textbox, added
support for shadow element with attributes, shadow color, shadow obscured )
b) use proper fillcolor attribute
c) detect whether note/comment is shown on import
d) export state of note ( shown/hidden )
text position still not perfect, we still are missing some attributes to control
that.
Change-Id: I59b446175217479ce7960287aa540df8c6b2b1ea
i.e. since b35980d9b
seems remove, remove, delete was old pattern, new patterns ended up
as remove+delete, remove. Reorder to remove, remove+delete
Change-Id: I54ec8d0296e751110c516516465be2ac0615f6a1
BTW, do we really need/want this (outdated) fork of DocumentLoader?
Either LibreOffice4Android should use the "real" DocumentLoader class
(and that should then be modified as necessary, but making sure it
still works as a separate test app), or this fork should be called
something else and be radically different.
Change-Id: I8536336140ee0190d01d35dde1a4c78e7b22c6bd
Until now, only the ToolBar's TextField stored the history, but
this is destructed when ToolBar is closed. The "Manager" class is alive
in case of Toolbar's invisibility, so I made this class to store history
temporarly.
Change-Id: I816dd3c83440000ab4e6caaeb2cccf390d3042fd
This is a preparation for gbuildizing dictionaries, which does not
contain registration/LICENSE like regular extensions.
Change-Id: I9b186d101627872b522be0432f6daf21250eb497
Put two experimental options (Experimental & Macro Recording) below
a linear divider with the header of "Optional (Unstable) Features"
Had to adjust distance between two options in order for them to fit
properly in the dialog.
Change-Id: I784f99c282d5b70e7e025b1ca05a7c8d73869cef
After a string has been passed as key parameter to hash_store, it must
not be modified, as happens in the print_nodotdot and print_fullpaths
functions; if it doesn't crash then the dep files are twice as large
in a MSVC build because duplicates are not eliminated.
(regression from c2b467b84a)
Change-Id: I0ae96a7fbcefa20b118717d923f60b4f2255a642
... actually only SvtFileDialogURLSelector has it.
The problem is that in this case ImplGetSeparatorX() returns 0.
Change-Id: I2f5bebce31b4f137782b5f34cc7954554af06863
No one hit upon an idea to hold the "Up One Level" button in the file picker
to notice this? :) Only the bug in MenuButton has helped to discover.
Change-Id: I760eff05dd19c3fdeb4b2052c0c20dbde6f6d2ed
...in loadImplementation (instead of using the context the ServiceManager itself
was created with). Otherwise, the handcrafted context containing a fake
theJavaVirtualMachine singleton in install_vm_singleton
(javaunohelper/source/vm.cxx) would not be honored, so that if a Java process
bootstraps native (binary) UNO and from there tries to obtain that singleton, it
would erroneously try to instantiate another JVM instead of using the existing
one. This was a regression introduced with the new ServiceManager and could be
witnessed by test-javanative in ure/source/uretest/Makefile failing.
Change-Id: I58cfbc8cdaea7ee4ab80fac728ea3e85676d69e1
...this had been broken with gbuild'ification of javaunohelper and caused
java.lang.UnsatisfiedLinkError from
com.sun.star.comp.helper.Bootstrap.cppuhelper_bootstrap, as could be witnessed
by test-javanative in ure/source/uretest/Makefile failing.
Change-Id: I8a76e1195c713895bfb8eae5070b0f73beb2b897
...at least, Makefile (to be run from within an SDK environment) works again; I
reflected all the relevant changes in Makefile.pln (to be run from no specific
environment) too, but did not actually check the latter
Change-Id: Ie2012d26b3bd59335a0f872bbfc1414cc4f5edc5
Regression from commit b3cee382, the bugdoc is an example of why those
can't be optimized with the current method. (Yes, it is somewhat
special, after a resave in Word, the bug is no longer there.)
Change-Id: I4b5e82b13242726d6df47c447642ca8916aa6f31
