Also make the --with-android-ndk-toolchain-version option mandatory. Otherwise
it would be possible to leave it out and that would appear to work in the
configure phase but in fact mix stuff up when using a self-built 64-bit
tool-chain merged into a r8b,c,d NDK.
Sigh, maybe we shouldn't even try to have the configury generic enough to work
with any NDK version (which probably it doesn't anyway), but explicitly accept
only a small set of NDKs, like r8b,c,d (which probably are the only that we
work with anyway).
Change-Id: Iecdf3a2d841e7b29598228ae49f48eca9640a82c
That means that there is corresponding .po file for given language.
As a slight optimization, do not even try to run uiex if _no_
translation would be produced.
Change-Id: I901e88b20edfb55e4f8bc661bacf8218b603bdcb
When running a unit test under GDB, we do not want to prefix the
cppunittester executable with the LD_LIBRARY_PATH, because that has
already happened before the "gdb --args" part of the command line.
Change-Id: If3f81ba3fc3e5260142d7e9c2d4a78e9ca63382c
Reviewed-on: https://gerrit.libreoffice.org/1532
Reviewed-by: Tor Lillqvist <tml@iki.fi>
Tested-by: Tor Lillqvist <tml@iki.fi>
The form: OuterCtor( InnerCtor(array[i]).method() )
with explicit ctor(s), unused optional arg on ctor+method
--> error 'parameter can not have variably modified type'
gcc version 4.4.3 (Ubuntu 4.4.3-4ubuntu5.1).
Min fix: force precedence/associativity in type resolution.
fdo#58417 BTs inside 1 of these ctor calls.
Change-Id: I4d8b979df0fc827758b1e9a336031671425486ea
Reviewed-on: https://gerrit.libreoffice.org/1531
Reviewed-by: Tor Lillqvist <tml@iki.fi>
Tested-by: Tor Lillqvist <tml@iki.fi>
OOXML spec says:
"[szCs] specifies the font size which shall be applied to all
complex script characters in the contents of this run when displayed"
Change-Id: I0faf599e38ef5a2e2005bb6f98874639be7d8287
Reviewed-on: https://gerrit.libreoffice.org/1454
Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>
All these are years old, guessing from some comments as old
as Windows 9x support. These workarounds do not seem to be needed now.
Change-Id: If67baa5cdefcec33b28696c764b1ed96143b7ccd
Load of the last Misc. config item is skipped due to
its index == # of strings in array in GetPropertyNames.
Omission in middle drops last item -
never gets to the Enable Macro Recorder option.
Change-Id: If99e0aa89e82ffc362f5e4bda712881a546e510a
Reviewed-on: https://gerrit.libreoffice.org/1523
Reviewed-by: Tor Lillqvist <tml@iki.fi>
Tested-by: Tor Lillqvist <tml@iki.fi>
I am constantly amazed at the creativity of the original makefile
writers. An extra header file, processed by sed, rather then adding one
item to CDEFS? Really?
Change-Id: I41ae8b10fc447ea5ab91e767c8afb87e39b9b5f5
The fields were not really unused but the compiler can't know that,
and it was done in a horribly hacky way anyway.
Same fix as sberg did for the corresponding MacOSX code. (Yeah, should
factor out the commonality. Not that I know whether the VCL code for
iOS will ever be used.)
Change-Id: I573073c3f5c15f0a40ed72c9d58578fc80f65b93
