I get exactly the same kind of artefacts as in the Android app, which
I guess is good as it is at least consistent, as the implementation at
the LO layer is identical...
Change-Id: Icf0690fd2c48a133cb66de2ab7977b7088d2199e
Cannot satisfy all platforms warning-free with one C format string
(sometimes it is "unsigned int" and sometimes "unsigned long"),
so avoid the problem altogether and use C++ streams.
Change-Id: I898d776c753dedbd79d4a56c580912613e865dda
This reverts commit c5e5699c80.
And adapts to OUString.
Conflicts:
basic/source/runtime/runtime.cxx
Change-Id: Icd7c1e1e57162eefb1f3631aa5509fd3a09c9b08
- nanosecond precision
- signed (allowed negative) year
Also: assorted improvements / bugfixes in date/time handling code.
Some factorisation of copy/pasted code.
Change-Id: I761a1b0b8731c82f19a0c37acbcf43d3c06d6cd6
Don't see how it could be used again, it appears to be leaked
and causes the assertion in PyThreadState_Swap to fire.
Change-Id: Id6bbb4363928e4fdd8bda1905f6e1f9931dba74c
Reviewed-on: https://gerrit.libreoffice.org/3452
Reviewed-by: David Ostrovsky <David.Ostrovsky@gmx.de>
Tested-by: David Ostrovsky <David.Ostrovsky@gmx.de>
some improvement to the docking area layout on a undock-dock-cycle of a window
(cherry picked from commit cad2fc1a82a3efb2abd4396fe2efc2231fb783e5)
Conflicts:
sfx2/source/dialog/splitwin.cxx
Change-Id: I39d888248a973304163bbe499094bef51a9d61f1
if we're in edit mode, SC_MOD()->InputEnterHandler() kicks
us out of it. But the very similar ScFormatShell::ExecuteAlignment
method doesn't do an InputEnterHandler. I can't see why we
would want to do this, and without it we get the desired
behaviour that tex direction changes and we remain in whatever
mode we were in before changing it
Change-Id: Iaf6e154ffb8f317fda997b302d4075d2efcac8db
Reviewed-on: https://gerrit.libreoffice.org/3450
Reviewed-by: Fridrich Strba <fridrich@documentfoundation.org>
Tested-by: Fridrich Strba <fridrich@documentfoundation.org>
Currently the copy function stores the paragraph attributes
to SvStream but the restoration in paste isn't implemented yet!
This patch implements the attributes loading from SvStream.
Change-Id: I58928ca66a83b3fe3c3894a28ccf1e7fe5f661b3
Reviewed-on: https://gerrit.libreoffice.org/3369
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>
* RTL_CONTEXT_foo is originally intended to be used for performance
measurement, but mostly unused right now
* however its macros are sprinkled all over the codebase and might
provide good help for ad-hoc debugging, and prevent SAL_INFO
duplication
* if there is no performance logging, these macros are now forwarded to
SAL_INFO with logarea "logfile" or "$AREA.logfile"
Change-Id: I495c8924e5b55124f7eac68eccda074d0740c117
Reviewed-on: https://gerrit.libreoffice.org/3421
Reviewed-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
Tested-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
This is somewhat non-obvious since there is no way to override curl
configure's use of pkg-config for NSS... it needs actually 2 include
paths, and the --with-nss takes only one (which is extended with
include/ and lib/).
Thanks to Alex Thurgood for the report.
Change-Id: I06074de50daad19a2cc62fe4090d3c0fbfdb2cb1
The mpServerFont member of a ImplServerFontEntry must not be deleted while the
ImplServerFontEntry still exists
see also 39cbce553d for another reason a crash in
the same place can happen. Its impossible from traces in crashes before
39cbce553d was fixed to distinguish those crashes
from this crash.
This crash is a regression due to 7a416820ab
where we went from modifying pServerFont in X11SalGraphics::setFont directly to
modifying it/a-different-one indirectly via ImplServerFontEntry
The various font caches and font thing lifecycles of LibreOffice are somewhat
confusing.
This crash had eluded me for years, to reproduce:
insert->special chars->select a font with loads of glyphs, i.e. "AR PL UKai CN"
click on the first row of glyphs and hold down page-down until you hit the
bottom, then page-up until you hit the top. Pre patch it won't survive the
whole down+up (and valgrind will moan quite a bit)
Change-Id: Ifde0cb375f487c556b04a640d77765a7dc2f0913
I guess all those $(subst \\,\,...) in odk/examples/.../Makefile are only
workarounds for a broken PS setting and could go, too.
Change-Id: I2e44108f23803e1f49157a91be11af5ad860673c
...otherwise preload.cxx in juh.dll will not be able to find them
via osl_getFunctionSymbol. What apparently happens is that JNICALL expanding
to __stdcall decorates symbol names with _...@NN. This likely was hidden in
pre-gbuild times thanks to the use of def files. (On a side note, the JVM
appears to contain special code to find syms for native methods in both
decorated and undecorated form; that explains why it picks up the decorated
symbols from juh.dll just fine.)
There is no need for the functions in juhx.dll (called from the juh.dll wrapper)
to adhere to JNICALL (in fact, things would likely be easier to maintain if the
juhx.dll functions also used different names than their juh.dll wrappers).
However, what complicates this patch is that for DISABLE_DYNLOADING, the juh
wrapper and its preload.cxx is elided, and the code that would normally go into
the juhx library goes into the juh library (and thus needs to stick to JNICALL,
and also needs to use the right function names).
Change-Id: I66611648f1f79f57f0c1b23fb7a801da2d7b86c5
fixes reconnect crash. Won't crash when server-end disconnect.
seperate Client construction and initialization so that Client()
will release its mutex lock and won't block service.run().
Otherwise onBackPressed() will be blocked in PairingActivity
Change-Id: I424a470aa02b0c74b28cb9f9ba79489aa0d4ab1b
