On --disable-openssl, the bundled neon library
will link against GNUTLS + gcrypt instead of
OpenSSL.
Change-Id: I5b3f09cd1003aefde0478aaab026536c962212c4
Reviewed-on: https://gerrit.libreoffice.org/3330
Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>
On --disable-openssl, let's avoid linking the
bundled postgresql to OpenSSL by not passing down
--with-openssl to its configure script.
Also, configure stage will fail if krb5 or gssapi
are enabled as they need OpenSSL and, in any
case, --with-krb5 and --with-gssapi will not be
passed down to postgresql configure script.
Change-Id: Iaf7e944d1d8f6a018e949ece56f6d3881f1e8c46
Reviewed-on: https://gerrit.libreoffice.org/3333
Tested-by: LibreOffice gerrit bot <gerrit@libreoffice.org>
Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>
Most of the components included in LibreOffice
already support alternative TLS libraries (e.g.
NSS, GnuTLS).
Change-Id: If00c348046fdbc88156f3d89c25e874e7e9bd04c
Reviewed-on: https://gerrit.libreoffice.org/3328
Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>
When expliciting openssl as tls implementation to
use, show a warning as there are still some pieces
of code that may depend on NSS or GNUTLS.
Change-Id: I2cfaaf09b5ad83837928f2648975db956b261c19
Reviewed-on: https://gerrit.libreoffice.org/3327
Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>
This is in preparation for removing gb_ZipPackage. It is only a
temporary workaround, as the URE headers can be used directly from
$(SRCDIR), once we do the migration of public headers to top-level.
This commit should be reverted then.
Change-Id: I26b7ad6eb7b33797c174573e3c2f69d295408657
The commit message of 0e5f2ea0de71439fe121da47deaba68585541808? didn't
mention this, so I guess flipping over the default was just an accident.
Change-Id: I9078cc56a6064e3498b012848b860c7b414b37e6
Support cramming everything into one big merged library as well as
only the libraries depended upon by almost everything.
Change-Id: I69647037dc62840294ba670d5d6c172a4608bf3f
There is no configure switch for this, URELIBS must be set.
This commit changes strategy to link also libraries being merged.
We need them for build tools like idlc, cppumaker, .., so the tools can
link against them now. This avoids circular dependencies.
Change-Id: Ic49e18ecbeaff84d4f5a7fafe8b1fbf45ed18c9b
one variable to find them,
one variable to deliver them all and into filelist put them,
in $INSTDIR where the installer searches.
Change-Id: I989f578f0ed6f9ef9167522249b36d95c15bfd1b
I think it is useful to use SAL_FINAL mainly as a documentation aid,
to make it clear to a code reader when a class is not expected to be
derived from, and when a virtual function is not expected to be
overridden in a derived class.
Possibly there is also some class of bugs that using SAL_FINAL will
help find?
Change-Id: I45002f020dcb52e8a9f2962ff98780f2b80627af
I get the below warning for every single jar file without this patch.
warning: [options] bootstrap class path not set in conjunction with
-source 1.5
1 warning
Change-Id: I71c01aeea993640f1ec86fe1d8a977656861358d
Since we no longer support the old Apple SDK using gcc-4.0.1, we can
remove the cruft to work around its problems. Woohoo.
Change-Id: Idf275e76449443f1f0314e75dab993f213a77eb7
upgrade raptor to 2.0.9, rasqal to 0.9.30, redland to 1.0.16:
- new raptor depends on libxml2 version 2.6.27 and thus will _not_
build against system libxml2 on Mac OS X 10.4/10.5
- stop delivering raptor-config/rasqal-config, set env vars instead
(which unfortunately requires patching configures, TODO: upstream that)
- remove redundant configure flags
- stop passing -Wl,no-inhibit-exec (causes HP-UX code to be built on Linux)
- disable new rasqal build deps:
--with-uuid-library=internal --with-digest-library=internal
- disable redland "libltdl" nonsense via --disable-modular
- fortunately openssl digest support has been removed too
- drop raptor-1.4.18.ooo_build.patch, set RAPTOR2_CFLAGS/LIBS instead
- drop raptor-1.4.18.autotools.patch:
+ FreeBSD parts moved to raptor-freebsd.patch.1
+ Android parts obsolete
+ don't know about the MinGW part (TODO)
- from drop raptor-1.4.18.mingw.patch dropped the part defining parsers
to be built, done via --enable-parsers
- drop redland/raptor/raptor-1.4.18.cross.patch:
the vsnprintf check is now done at runtime, and expat has been removed
- drop raptor-1.4.18.aix.patch,redland-1.0.8.aix.patch:
presumably don't need special check for V7BETA since it's released now
- drop raptor-1.4.18.rindex.patch, index/rindex no longer used
- drop raptor-1.4.18.entities.patch, fixed upstream in 2.0.7
- drop raptor-1.4.18.fixes.patch, fixed upstream in 2.0.9
- drop raptor-1.4.18.{libxml,libxslt}.patch: TODO
- drop {raptor-1.4.18,rasqal-0.9.16,redland-1.0.8}.legal.patch:
we don't care any more about Sun legal dept.
- drop rasqal-0.9.16.ooo_build.patch, set RASQAL_CFLAGS/LIBS instead
- drop rasqal-0.9.16.autotools.patch:
+ FreeBSD parts moved to rasqal-freebsd.patch.1
+ Android parts obsolete
+ MinGW part only for cygwin build
- drop redland-1.0.8.autotools.patch:
+ FreeBSD parts moved to redland-freebsd.patch.1
+ Android parts obsolete
+ don't know about the MinGW part (TODO)
- drop redland-1.0.8.ooo_build.patch,
hopefully not necessary to disable utils build
Change-Id: I9cc44928ba87392d69e809abe2236cec0e4a508a
Reviewed-on: https://gerrit.libreoffice.org/3154
Tested-by: LibreOffice gerrit bot <gerrit@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
This reverts commit 151abb8b2b.
Unfortunately has the side effect that it will prevent GCC from
generating makefile dependencies for headers from bundled external
libraries, which breaks incremental builds horribly.
(Retain the uses in configure for real system headers).
Conflicts:
RepositoryExternal.mk
configure.ac
Change-Id: I149db1d402fa18bdc470f90dee846cfb5158237e
It is actually only the shell/Library_losessioninstall.mk that requires
GVariant, and that one is Linux-only, all other GIO stuff is fine.
Which means now e.g. the ARM tinderbox fails. Revert, and either ignore
or wait for somebody to write a better check.
This reverts commit e511402738.
This should be completely equivalent to ./autogen.sh CC=clang CXX=clang++ ,
which is the normal way to pass any compiler to configure, so this is needless
duplication.
This reverts commits 1a843cc54d and
02ed260819 .
Change-Id: Ib02ab84a7b84fe4797dca0f62aa5049c1be55302
Drop HAVE_FEATURE_HELP. Just check HAVE_FEATURE_DESKTOP instead in the few
places in the code where HAVE_FEATURE_HELP had been introduced so far. It's
only for the non-desktop platforms that leaving out the traditional help
functionality is intended anyway.
Change-Id: I2d05869267f509a8e64cf402eed8ebef0020cada
I started running into mysterious trouble when some code still thought it
should be looking for extensions, although (perhaps) other code was not
willing to support such, or something. Some confusion caused by my recent
changes to make the conditional support of features more fine-grained.
Change-Id: I61758cb06054ceb038d200bc95615b5b85ed0f7b
Defines HAVE_FEATURE_{DESKTOP,HELP,MULTIUSER_ENVIRONMENT} for
now. Using <config_features.h> will supersede checking the
LIBO_FEATURE_DESKTOP and LIBO_FEATURE_HELP passed on the compilation
command-line.
Change-Id: I9866e2fd142f5c1a64c8f5c74452a20de23c66e3
When running module-deps.pl postgresql gets built just so that
libpq-flags.mk can be included. Since we already have all the necessary
libraries, add them explicitly and avoid this.
Change-Id: Icd94fc215ecb26c95f9ae3c14625bf819bf3c5c3
