13
0
livetrax/libs/sigc++2/ChangeLog
Taybin Rutkin f76fc34c29 Removed libglademm.
Updated libsigc++ from 2.0.6 to 2.0.17.


git-svn-id: svn://localhost/ardour2/branches/2.1-staging@1413 d708f5d6-7413-0410-9779-e7cbd77b26cf
2007-02-03 02:13:18 +00:00

1851 lines
70 KiB
Plaintext

2.0.17:
2005-12-20 Murray Cumming <murrayc@murrayc.com>
* sigc++/macros/signal.h.m4: Make all operator--() methods
return by reference, like the operator++() methods.
Bug #304402 from John Profic.
This is the libsigc-2-0 branch, for maintenance of libsigc++ 2.0.
API/ABI-compatible API-addition happens in the HEAD branch.
2005-12-01 Murray Cumming <murrayc@murrayc.com>
* sigc++/functors/slot_base.cc:
slot_base::disconnect(): Set call_ to 0,
to invalidate the slot, even if parent_ is 0.
I think parent_ is, for instance, a signal, but
disconnect should still work on a slot that is not
connected to a signal, because a slot can be invoked
directly.
Fixes bug #311057 from James Lin.
2005-12-01 Murray Cumming <murrayc@murrayc.com>
* tests/Makefile.am:
* tests/test_slot_disconnect.cc: Added test
case from bug #311057.
2005-11-16 Philipp Berndt <philipp.berndt@gmx.net>
* sigc++/adaptors/macros/exception_catch.h.m4: Make member
exception_catch_functor<T_functor, T_catcher, void>::catcher_
public so that it can be accessed by visit_each()
(bug fixed for generalization on 2004-11-06)
2.0.16:
2005-08-01 Neal E. Coombes <nealc@trdlnk.com>
* sigc++/signal_base.h: Updated the documentation for temp_slot_list
as requested in bug #303896.
2005-08-01 Murray Cumming <murrayc@murrayc.com>
* sigc++/adaptors/hide.h.m4: Added missing
) in call to sun_forte_workaround(), fixing
build on SUN Forte 5.5. Bug #312020.
2005-08-19 Bruno Martinez <brunom@fing.edu.uy>
* sigc++/type_traits.h: Renamed
::sigc::is_base_and_derived::internal to
::sigc::is_base_and_derived::internal_class
in order to avoid conflict with
namespace internal.
2005-07-13 Murray Cumming <murrayc@murrayc.com>
* docs/manual/libsigc_manual.xml: Correct mentions of
1.2 stuff instead of 2.0. Patch in bug #310213 from
pebble.org.uk.
2005-07-13 Murray Cumming <murrayc@murrayc.com>
* docs/manual/libsigc_manual.xml: Fixed typo
found by Antonio Coralles.
2005-07-09 Murray Cumming <murrayc@murrayc.com>
* sigc++/macros/signal.h.m4: Did the same (see
last commit) for slot_const_iterator and
slot_iterator_buf.
2005-07-09 Murray Cumming <murrayc@murrayc.com>
* sigc++/macros/signal.h.m4: slot_iterator:
operator--() now returns value, not reference, like
operator++() already did. This caused crashes when
using --no-inline with g++. Bug #308651 by
Michael Andres.
2.0.15:
2005-07-04 Philip Langdale <plangdale@vmware.com>
* sigc++/adaptors/macros/compose.h.m4: Add a setter typedef to
compose*_functor and use it instead of the (incorrect) getter
typedef in the compose* specialization of visit_each<>().
This corrects the lifetime management of slots created with
compose(). Bug #308433.
2005-06-13 Marek Rouchal <marek.rouchal@infineon.com>
* tests/test_deduce_result_type.cc: Specify int return type
for main(), to be more ISO C++ compliant. Bug #307478.
2005-06-11 Andris Pavenis <pavenis@latnet.lv>
* sigc++/adaptors/lambda/macros/base.h.m4:
* sigc++/adaptors/lambda/macros/select.h.m4
* sigc++/adaptors/macros/hide.h.m4: Specify only a type (not a
parameter name) for unused member function parameters
2005-06-12 Paul Pogonyshev <pogonyshev@gmx.net>
* configure.ac:
* scripts/cxx.m4:
* sigc++config.h.in: Add test for whether the compiler allows
referencing to member functions of the class/structure being
declared from a definition of a static member variable.
Supposedly a generic solution for GCC 3.2 compilation problems.
* sigc++/type_traits.h: Define SIGC_WRAP_IS_BASE_CLASS_ based on
results of the above test.
(struct is_base_and_derived): Wrap up is_base_class_() functions
in an internal class if SIGC_WRAP_IS_BASE_CLASS_ is defined.
2005-06-10 Murray Cumming <murrayc@murrayc.com>
* sigc++/adaptors/macros/bind.h.m4:
* sigc++/functors/macros/slot.h.m4:
* sigc++/macros/signal.h.m4: Use CALL_SIZE instead of
hard-coded 7s and 6s.
2.0.14:
2005-06-10 Murray Cumming <murrayc@murrayc.com>
* sigc++/visit_each.h: Make the
limit_derived_target::with_type inner class an outer class,
to satisfy the SUN CC 5.7 compiler, though I think it is a
compiler bug. Bug #302098 has the test case.
2.0.13:
2005-06-07 Murray Cumming <murrayc@murrayc.com>
* tests/test_compatibility.cc: Specify the actual class when
using test_int(), instead of the derived class, to fix the build
on SUN Forte CC 5.5. Patch from Friedemann Kleint in
Bug #305647
2005-06-07 Murray Cumming <murrayc@murrayc.com>
* sigc++/macros/signal.h.m4: signal_emit::emit(): Use scope to
ensure a certain order of destruction of the member variables, to
avoid a leak on MSVC++. Patch by Andreas Ames in Bug #306249.
2005-06-07 Murray Cumming <murrayc@murrayc.com>
* sigc++/macros/signal.h.m4: Added comments about commenting-out
SIGC_TYPEDEF_REDEFINE_ALLOWED when using SUN Forte CC 5.7,
because I can not seem to create a test for it.
2005-06-07 Murray Cumming <murrayc@murrayc.com>
* configure.ac:
* scripts/cxx_std.m4:
* sigc++/macros/signal.h.m4:
* sigc++config.h.in: Added check for
the non-standard SUN Forte reverse_iterator<>,
and used it. This is based on the same stuff in
gtkmm.
2005-06-07 Murray Cumming <murrayc@murrayc.com>
* sigc++/visit_each.h: limit_derived_target(): Just some whitespace
changes.
2005-05-16 Neal E. Coombes <nealc@trdlnk.com>
* sigc++/signal_base.h: Modified temp_slot_list to be a temporary view
into a slot list. Instead of emptying the original it now simply tacks
a placeholder to the end of the original. It then uses this as it's
'end' iterator. This should allow for conscious recursiveness, as well
as inserting a slot to any position in the slot list during emittion.
See bug #303896.
2005-06-04 Friedemann Kleint <kleint@bifab.de>
* sigc++/macros/limit_reference.h.m4:
visit_each() template specializations:
Mention the bool I_derives_trackable
template type, to fix the build on Solaris
Forte 5.5.
2.0.12:
2005-05-06 Régis Duchesne <hpreg@vmware.com>
* sigc++/macros/limit_reference.h.m4 (added):
* sigc++/Makefile.am:
New class that just stores a reference, and makes sure that if the
reference derives from trackable, then the trackable reference will be
used instead of the derived reference in visit_each().
* sigc++/functors/macros/mem_fun.h.m4: Better fix for bug #169225 by
Régis Duchesne and Christian Hammond, based on the new limit_reference
class.
* sigc++/adaptors/bound_argument.h (added): New class (built upon the
new limit_reference class) that handles all 3 kinds of bound arguments:
by value, by reference, and by constant reference. References are
unwrapped in the bound_argument's constructor.
* sigc++/adaptors/macros/bind.h.m4: Fix for bug #302327 by Régis
Duchesne. Bound arguments now need to know whether they are passed by
reference or not. So bind() now builds bind_functor instances using
'reference_wrapper<Foo>' types, instead of 'Foo &' types. The
bind_functor code is modified to compensate.
* sigc++/adaptors/macros/bind_return.h.m4: Similar fix for a similar
bug (unfiled) with bound return arguments.
* sigc++/reference_wrapper.h:
* sigc++/type_traits.h:
The reference_wrapper class is only used in bound_argument.h. Put
correct but unused code under #if 0.
* sigc++/adaptors/lambda/base.h: This file needs reference_wrapper.h,
but was incorrectly relying on type_traits.h to include it.
* tests/Makefile.am:
* tests/test_virtualbase_delete.cc (deleted):
* tests/test_virtualbase_delete_ref_param.cc (deleted):
* tests/test_limit_reference.cc (added):
Replaced test_virtualbase_delete*.cc with a simpler
test_limit_reference.cc which checks for all 3 aspects of the same bug
in one file.
* tests/test_bind_ref.cc: Slots must use 'Foo &' types. We were lucky
this broken usage worked before this change. The change in
type_traits.h made this bug obvious, by preventing the code to compile.
* tests/test_bind_return.cc: After my change, bind() and bind_return()
must use 'reference_wrapper<Foo>' types.
* tests/test_custom.cc: Made this test a no-op (so it does not perturb
'make check' on released versions of the library) and made it a
template ready to be modified by hackers.
2005-05-01 Murray Cumming <murrayc@murrayc.com>
* sigc++/functors/slot_base.cc:
slot_base::slot_base(src): If the source
slot_base has a null rep->call_, meaning that the
slot is invalid, just return a default-constructed
slot, to prevent the crash shown in
tests/tests_copy_invalid_slot.cc. Bug #302515 by
Régis Duchesne.
2005-05-01 Murray Cumming <murrayc@murrayc.com>
* sigc++/functors/macros/mem_fun.h.m4: bound_*<>:
Add a new is_base_and_derived<sigc::trackable>
parameter to the template and thereby provide a
specialization for T_Obj types that derive from
sigc::trackable. This prevents a crash when casting
from the derived type to sigc::trackable after the
derived destructor has run. This cast can sometimes
fail when using multiple inheritance, at least with
g++. Bug #169225 by Régis Duchesne and Christian
Hammond.
* sigc++/type_traits.h: Add documenation for
the internal is_base_and_derived<> template, which
allows us to specialize other templates for certain
template types.
2005-04-28 Murray Cumming <murrayc@murrayc.com>
* sigc++/type_traits.h: Added comments for users of
g++ 3.2.
2005-04-28 Murray Cumming <murrayc@murrayc.com>
* tests/Makefile.am:
* tests/test_virtualbase_delete.cc: Added
simplified test case from bug #169225. We have a patch
to make this succeed but I am not ready to commit it
just yet.
2005-04-27 Murray Cumming <murrayc@murrayc.com>
* tests/Makefile.am:
* tests/test_custom.cc:
Added a place to put extra test code, so I don't have
to keep installing my crazy libsigc++ versions.
2005-04-27 Murray Cumming <murrayc@murrayc.com>
* sigc++/visit_each.h: Revert back to the non-explicit
template call, because we can not specify the
template specialization so easily, because the
specializations have different numbers of types.
* tests/Makefile/am:
* tests/test_bind_ref.cc: Add a simple test only for
sigc::ref disconnection.
2005-04-26 Murray Cumming <murrayc@murrayc.com>
* sigc++/visit_each.h: Use the explicit template
specialization, needed for Tru64 and AIX compilers.
This causes a crash in some uses of sigc::ref()
(when using g++ 3.3.4 or 3.3.5 , but not with 3.4) but
seems to fix a crash in some uses of multiple inheritance
(bug #169225).
* tests/test_bind.cc: Comment out the crashing (not with
g++ 3.4) use of sigc::ref() with an explanatory comment.
2.0.11:
2005-03-09 Cedric Gustin <cedric.gustin@swing.be>
* Makefile.am: Moved MSVC_Net2003 directory to SUBDIRS.
* MSVC_Net2003/Makefile.am: Added blank.cpp to EXTRA_DIST.
* MSVC_Net2003/libsigc++2_msvcNet2003.sln: Removed
test_accumulated and test_lambda projects as the tests are
disabled in the standard, configure-based build.
* MSVC_Net2003/MSVC_Net2003/libsigc++2_msvcNet2003.vcproj:
Generate a PDB file in the Debug target.
2005-03-05 Murray Cumming <murrayc@murrayc.com>
* docs/website/stable.html: Mention actual supported compilers.
* docs/website/docs.html: Removed broken examples links. Change
reference link to the reference overview page.
2005-02-23 Murray Cumming <murrayc@murrayc.com>
* sigc++/functors/slot_base.h: Make slot_base::operator=() public, to
fix the build with SUN Forte C++ 5.5 and Tru64.
2.0.10:
2005-02-20 Murray Cumming <murrayc@murrayc.com>
* tests/test_slot.cc, test_disconnect.cc: #included <new> to avoid
an unresolved symbol error with the Tru64 compiler. Solution found by
Tim Mooney in bug #161503.
2005-02-20 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/signal_base.h: Add some documentation.
2005-02-20 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/signal_base.cc: Reset deferred_ flag to false in
signal_impl::sweep() (Neal E. Coombes). Partly fixes bug #167714.
2005-02-11 Martin Schulze <mschulze@cvs.gnome.org>
* docs/manual/Makefile.am: Set the correct group in post-html.
2005-02-11 Murray Cumming <murrayc@murrayc.com>
* docs/website/doc.shtml: Fix typo in url for reference docs. Found by
James Lin.
2005-02-06 Murray Cumming <murrayc@murrayc.com>
* sigc++/signal_base.h: temp_slot_list::begin(), end(): Actually
return the iterators.
2005-02-03 Neal E. Coombes <nealc@trdlnk.com>
* sigc++/signal_base.h: Add temp_slot_list struct to facilitate
allowing new connections to a signal during an emittion without
affecting that emittion.
* sigc++/macros/signal.h.m4: Use the new struct temp_slot_list to
prevent connections made during an emittion from being called in the
same emittion (which has the potential of causing things like
infinite loops). This guarantees an emittion will be finite, as well
as maintaining any order of emittion guarantees that may have already
been in place.
2.0.9:
2005-02-02 Murray Cumming <murrayc@murrayc.com>
* sigc++/visit_each.h: visit_each_type(): Reverted the change, so that
we do not specify the template types ot visit_each<>(), because it
crashes on g++ 3.3.4 (but not g++ 3.4.2). Added a comment telling users
of AIX (and maybe IRIX MipsPro and Tru64) to use the alternative version
if they have compilation problems.
2005-02-01 Murray Cumming <murrayc@murrayc.com>
* sigc++/adapators/macros/base.h.m4: Add sigc::var<>() documentation,
from Roger Ferrer Ibáñez in bug #149483.
2005-02-01 Murray Cumming <murrayc@murrayc.com>
* sigc++/adaptors/macros/compose.h.m4: In the template specializations
of visit_each(): As before, specify the specific other template
specializations of visit_each that we use. Needed by AIX.
* tests/Makefile.am: Reenabled most tests, because AIX can now
build them.
2005-02-01 Murray Cumming <murrayc@murrayc.com>
* sigc++/visit_each.h: visit_each_type(): Specify the specific
template specialization of visit_each<>() to use. The AIX compiler,
and possibly the Tru64 compiler, need this extra hint.
2005-02-01 Murray Cumming <murrayc@murrayc.com>
* bind.h.m4: Define bind_functor::operator() inline because the AIX
compiler/linker sometimes fails to find it when it is defined
outside of the class.
2.0.8:
2005-01-30 Murray Cumming <murrayc@murrayc.com>
* sigc++/type_traits.h: is_base_and_derived: Added
avoid_gcc3_warning_(), to avoid an incorrect warning when using
g++ 3.3.5
2005-01-28 Liza Klerck <liza@trdlnk.com>
* sigc++/functors/macros/mem_fun.h.m4: Add a sigc:: namespace
prefix to the nested use of visit_each(), to avoid ambiguity when
using 2 versions of the libsigc++ API inside different namespace.
2005-01-27 Murray Cumming <murrayc@murrayc.com>
* sigc++/adaptors/macros/adaptor_trait.h.m4: Add a sigc:: namespace
prefix to the nested use of visit_each(), to avoid ambiguity when
using 2 versions of the libsigc++ API inside different namespace,
which is not very advisable anyway. Bug #165222 from
liza at trdlnk.com.
2.0.7:
2005-01-24 Cedric Gustin <cedric.gustin@swing.be>
* sigc++config.h.in : Moved the SIGC_USING_STD macro definition
out of the SIGC_CONFIGURE section. We also need it for MSVC.
* MSVC_Net2003/.cvsignore : Added .def and .aps files
* MSVC_Net2003/Makefile.am: Rewrote rule for local copy of
sigc++config.h (required for 'make distcheck').
* MSVC_Net2003/libsigc++2_msvcNet2003.sln: Added test_retype and
test_disconnect_during_emit tests.
* MSVC_Net2003/tests/test_disconnect_during_emit/*,
MSVC_Net2003/tests/test_retype/*: Initial commit.
2005-01-21 Murray Cumming <murrayc@murrayc.com>
* tests/: Disabled the test_accumulator, test_bind, and test_compose
tests, and part of test_mem_fun because the AIX xlC compiler can not
build them, but it can still do most things, including the examples.
See the comments in tests/Makefile.am.
2005-01-21 Murray Cumming <murrayc@murrayc.com>
* sigc++/adaptors/bind.h.m4: non-member operator()(): Specify the
extra nil arguments in the templated class name prefix. Oddly, the
AIX xlC compiler says that the type of the first parameter does not
match the template if you don't do this.
2005-01-21 Murray Cumming <murrayc@murrayc.com>
* sigc++/type_traits.h: is_base_and_derived struct: Move the
is_base_class_() functions out of the inner class, because the AIX
xlC compiler does not like that - see the comments in the code.
* sigc++/adaptors/bind.h.m4: Add the extra nil template types to
the template specializations, as in slot and signal.
2005-01-21 Murray Cumming <murrayc@murrayc.com>
* sigc++/functors/macros/slot.h.m4, sigc++/macros/signal.h.m4:
slot and signal template specialization for
various numbers of template args: In the class slot line, specify
all the remaining template types as null, instead of expecting the
compiler to guess them in itself. This partly fixes the build on
AIX with the xlC compiler. Bug #164685.
2005-01-19 Murray Cumming <murrayc@murrayc.com>
* sigc++/type_traits: struct is_base_and_derived: Make the test inner
struct a friend, so that it can use the big inner struct. This is
required by the Tru64 compiler.
* sigc++/adaptors/lambda/base.h: Put the unwrap_lambda_value()
definitions at the top, because Tru64 (understandably) needs them to
be declared before use.
2005-01-19 Murray Cumming <murrayc@murrayc.com>
* scripts/: Added cxx_std.m4, with a test copied from
glibmm/scripts/cxx_std.m4 to check if the compiler has the std::
namespace.
* sigcconfig.h: #undef the new #define and add SIGC_USING_STD(),
like GLIBMM_USING_STD, to put stuff in the std:: namespace when it
is not there already,
* configure.in: Used the new test.
* tests/*: Uses SIG_USING_STD() for every std:: thing that we use.
This is needed by the Tru64 and HP-UX compilers when using their
defaults.
2005-01-19 Murray Cumming <murrayc@murrayc.com>
* configure.in: AC_INIT(): Provide the extra tarball name parameter,
so that it does not create a libsigc--- tarball.
2005-01-19 Murray Cumming <murrayc@murrayc.com>
* configure.in: AC_INIT(): Use libsigc++ instead of sigc++, attempting
to get the correct tarball name.
2005-01-18 Murray Cumming <murrayc@murrayc.com>
* configure.in: Used the autoconf 2.93 and AM_INIT_AUTOMAKE()
technique to specify ustar format for the tarball, to prevent files
with long file names from appearing at the top of the tarball.
Based on the same fix in gtkmm 2.6.
2005-01-18 Murray Cumming <murrayc@murrayc.com>
* sigc++/functors/macros/slot_h.m4: Specify the base class when
using the rep_ member variable. This stops the HP-UX aCC compiler
from saying that a Nonstatic member is referenced in a nested class,
local class or static member initializer. Bug #150719.
2005-01-18 Murray Cumming <murrayc@murrayc.com>
* Bug #159597 - patch from e97_far at e.kth.se to replace C-style
casts with reinterpret_cast<> and static_cast<> to avoid warnings.
2005-01-17 Murray Cumming <murrayc@murrayc.com>
* docs/manual/Makefile.am: Specifying html/index.html instead of
just the html directory as a target seems to fix distcheck problems.
I can also now confirm that the install works on solaris, when using
gmake, though not when using make.
2005-01-17 Murray Cumming <murrayc@murrayc.com>
* MSVC_Net2004/Makefile.am: Add built files to DISTCLEANFILES to fix
the distcheck.
* docs/reference/Makefile.am, manual/Makefile.am: Specify $srcdir in
paths, to fix distcheck of the manual, and maybe fix install problems
on Solaris.
2005-01-11 Murray Cumming <murrayc@murrayc.com>
* docs/website/stable.html: Updated the text about binary packages.
* docs/website/docs.html: Link to the 2.0 documentation instead of the
1.2 documentation.
2004-12-17 GregSchussman <schussman@slac.stanford.edu>
* glossary.shtml: Clean up punctuation, make definitions complete
sentences, and add clarifications for certain definitions according to
what Murray Cumming's suggestions and answers to my questions.
Bug #161580.
2005-01-11 Murray Cumming <murrayc@murrayc.com>
* docs/: Added manual, copied from the libsigc++-1.2 cvs module,
and updated it for the new 2.0 API.
2005-01-11 Murray Cumming <murrayc@murrayc.com>
* docs/: Added website, copied from the libsigc++-1.2 cvs module.
We will use it from here from now on.
2004-12-11 Cedric Gustin <cedric.gustin@swing.be>
* configure.ac : parse version tags at configure time (for
sigc-2.0.rc).
* MSVC_Net2003/sigc-2.0.rc.in : New resource file.
* MSVC_Net2003/Makefile.am: include sigc-2.0.rc in distribution.
2004-12-08 Cedric Gustin <cedric.gustin@swing.be>
* MSVC_Net2003/Makefile.am: get sigc++config.h from $(top_builddir)
instead of $(top_srcdir).
2004-12-08 Cedric Gustin <cedric.gustin@swing.be>
* MSVC_Net2003/*/*.vcproj: Renamed libsigc++ target to
sigc-2.0d.dll (Debug) and sigc-2.0.dll (Release). Added
$(SolutionDir) and $(SolutionDir)\.. to "Additional Include
Directories" in tests projects.
* sigc++config.h.in: Rewrote dllexport/dllimport macros for
MSVC, for better consistency with glibmm/gtkmm.
* MSVC_Net2003/Makefile.am: copy sigc++config.h from $(top_srcdir)
at build time.
2004-11-27 Murray Cumming <murrayc@murrayc.com>
* configure.in: Revert the AC_PROG_LIBTOOL change, so that this builds
with actually released libtool versions, and in jhbuild, so that it
gets testing.
2004-11-06 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/adaptors/macros/exception_catch.h.m4: Make catcher_ member
public so that it can be accessed by visit_each() (bug reported on
ml by Philip Langdale <plangdale@vmware.com>).
2004-10-24 Martin Schulze <mschulze@cvs.gnome.org>
* MSVC_Net2003/*/*.vcproj: Link with the "multithreaded DLL" runtime
libraries and enable RTTI for the MSVC build
(patch from Timothy M. Shead <tshead@k-3d.com>).
* MSVC_Net2003/*/.cvsignore: Hide generated build files from cvs
(patch from Timothy M. Shead <tshead@k-3d.com>).
2.0.6:
2004-10-12 Martin Schulze <mschulze@cvs.gnome.org>
* MSVC_Net2003/*/*.vcproj, MSVC_Net2003/blank.cpp: Fix project files
to compile out-of-the-box and add dummy file so that .cc files get
recognized as c++ code files (patch from Timothy M. Shead).
2004-10-10 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/signal_base.{h,cc}, sigc++/functors/slot_base.{h,cc},
sigc++/functors/macros/slot.h.m4: If SIGC_NEW_DELETE_IN_LIBRARY_ONLY
is defined, implement signal_base::operator new/delete and
slot_rep::operator new/delete (suggested by Timothy M. Shead).
Remove old work-around from 2004-10-02 since it didn't work.
2004-10-07 Martin Schulze <mschulze@cvs.gnome.org>
* configure.ac: Update for libtool 1.5a (with support for Intel C++).
* MSVC_Net2003/sigc++config.h: Remove bogus '#define'
(reported by Timothy M. Shead <tshead@k-3d.com>).
2004-10-02 Martin Schulze <mschulze@cvs.gnome.org>
* configure.ac: Bump version number to 2.0.6.
* NEWS: Add ChangeLog summary for version 2.0.6.
2004-10-02 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/functors/slot_base.{h,cc}, sigc++/functors/macros/slot.h.m4:
Rename (typed_)slot_rep::detach to (typed_)slot_rep::destroy.
Call the dtor of the functor stored in typed_slot_rep from destroy().
A cleaner solution would be to add an additional "virtual" function
that calls 'delete' or a real virtual dtor. However, this would be
less efficient and might break the ABI. (Fixes #152323.)
2004-10-02 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++config.h.in, MSVC_Net2003/sigc++config.h,
sigc++/signal_base.cc, sigc++/functors/slot_base.{h,cc},
sigc++/functors/macros/slot.h.m4: Define and use new macro
SIGC_NEW_DELETE_IN_LIBRARY_ONLY to ABI-compatibly move
all calls to new and delete into non-inline library code.
2004-09-26 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/adaptors/lambda/macros/group.h.m4: Add a missing
template keyword in the definition of deduce_result_type::type
(hopefully fixes #152327).
2004-09-26 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/macros/object_slot.h.m4: Use correct bound_mem_functor
variants for const (volatile) methods (fixes #148744).
2004-09-01 Martin Schulze <mschulze@cvs.gnome.org>
* docs/index.html: Correct link to lambda module.
2004-09-01 Martin Schulze <mschulze@cvs.gnome.org>
* README: Update compatibility section.
2.0.5:
2004-09-01 Martin Schulze <mschulze@cvs.gnome.org>
* MSVC_Net2003/Makefile.am: Add sigc++config.h to EXTRA_DIST.
* configure.ac: Bump version number to 2.0.5.
* NEWS: Add ChangeLog summary for version 2.0.5.
2.0.4:
2004-08-21 Martin Schulze <mschulze@cvs.gnome.org>
* tests/test_lambda.cc: Use sigc::var("\n") instead of sigc::ref("\n").
Comment out the affected lines, nevertheless.
Sun FORTE and Compaq C++ can handle neither sigc::ref("\n") nor
sigc::var("\n"). I see more chances fixing sigc::var("\n").
* sigc++/adaptors/lambda/macros/base.h.m4: Add a comment about a
possible work around for sigc::var("\n") compiler problems.
* tests/test_compatibility.cc: Remove a 'const' keyword that prevents
the test case from compiling with the Sun FORTE.
* tests/test_trackable.cc: Remove a 'virtual' keyword and an unused
variable to avoid compiler warnings.
* NEWS: Add ChangeLog summary for version 2.0.4.
2004-08-03 Martin Schulze <mschulze@cvs.gnome.org>
* scripts/cxx.m4, sigc++config.h.in, configure.ac,
sigc++/adaptors/lambda/macros/operator.h.m4, tests/test_lambda.cc:
Rollback change from 2004-07-15: configure check
SIGC_OPERATOR_OVERLOAD_AMBIGUITY is not needed - the overload
ambiguity doesn't occur if the lambda operators take all arguments
as const reference.
* configure.ac: Bump version number to 2.0.4.
2004-08-03 James Lin <slimjimmy@mail.com>
* Added SIGC_API qualifier to all externally-visible non-template
classes/structs.
* Added #include <sigc++config.h> to the files that use SIGC_API.
* Added empty SIGC_API definition to sigc++config.h.in for non-MSVC
compilers. I'm not sure if this is the right place to put this
(probably not).
* Added MSVC-specific sigc++config.h to the MSVC project directory.
(The comment in it probably should be edited.)
* Changed MSVC project settings to output a multi-threaded DLL, set
the include paths to work (hopefully) out-of-the-box. Disabled
precompiled headers, since they just complicate things and
shouldn't be necessary for such a relatively project.
2004-08-01 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/type_traits.h: Remove type_trait<>::instance()
(was unimplemented and unused; caused problems with the MSVC).
2004-07-23 Martin Schulze <mschulze@cvs.gnome.org>
* Makefile.am: Fix typo concerning distribution of libsigc++-2.0.spec.
* AUTHORS: Mention our contributors for platforms Sun FORTE and Intel C++.
2004-07-15 Martin Schulze <mschulze@cvs.gnome.org>
* *.h.m4: Don't call operator()() in sun_forte_workaround(); rather copy
operator()(). Calling operator()() makes a copy of the arguments causing
wrong results if an argument type is a reference. Hopefully fixes #147311.
2004-07-15 Martin Schulze <mschulze@cvs.gnome.org>
* tests/test_lambda.cc: Break "std::cout << [expr] << a << std::endl;"
into "std::cout << [expr]; std::cout << a << std::endl;".
I hope this fixes #147313 where the right values for "[expr]" but wrong
values for "a" were written to std::cout for some compiler with optimizations
turned off.
2004-07-15 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/adaptors/lambda/macros/operator.h.m4: Correct return type deduction
of lambda expressions in lambda_operator*::operator()(). Might be related to
bug #147313.
* sigc++/adaptors/lambda/macros/group.h.m4: Use m4 macro _P_().
2004-07-15 Martin Schulze <mschulze@cvs.gnome.org>
* scripts/cxx.m4, sigc++config.h.in, configure.ac,
sigc++/adaptors/lambda/macros/operator.h.m4, tests/test_lambda.cc:
Add configure check SIGC_OPERATOR_OVERLOAD_AMBIGUITY for a SUN FORTE
compiler problem (bug #147391). Use it to decide whether the lambda
action operators may be overloaded (not doing so restricts the API slightly).
* sigc++/adaptors/lambda/macros/operator.h.m4: Add some doxygen comments
and remove attic code.
* sigc++/adaptors/lambda/macros/base.h.m4:
Add templates unwrap_lambda_type and unwrap_lambda_value() to support
the non-overloaded lambda action operators. Also add some doxygen comments
and remove attic code.
* sigc++/adaptors/lambda/macros/group.h.m4: Fix a bug that resulted in
gargabe values being passed on to the functor contained in the group adaptor
(partly fixes #147313).
2004-07-11 Martin Schulze <mschulze@cvs.gnome.org>
* scripts/cxx.m4, sigc++config.h.in, configure.ac, *.h.m4:
Split SIGC_CXX_TEMPLATE_SPECIALIZATION_OPERATOR_OVERLOAD
into SIGC_CXX_GCC_TEMPLATE_SPECIALIZATION_OPERATOR_OVERLOAD
and SIGC_CXX_MSVC_TEMPLATE_SPECIALIZATION_OPERATOR_OVERLOAD.
Remove LIBSIGC_TEMPLATE_PREFIX. Add template keyword to
SIGC_WORKAROUND_OPERATOR_PARENTHESES depending on the configure checks.
Should fix the compiler problems with MSVC.
2004-07-11 Martin Schulze <mschulze@cvs.gnome.org>
* examples/hello_world.cc: Use sigc::ptr_fun instead of std::ptr_fun.
(fixes bug #144846)
2004-07-11 Eric Bourque <ericb@computer.org>
* libsigc++-2.0.spec.in: new file
* configure.ac : patched generate spec file
* .cvsignore: ignore generated file (Martin Schulze)
* Makefile.am: distribute spec file (Martin Schulze)
2004-07-11 Murray Cumming <murrayc@murrayc.com>
* sigc++/connection.cc: Added some comments.
* sigc++/trackable.cc: operator=(): Check for self-asignment, though I
do not know of any actual bug that this fixes. Added some comments.
* sigc++/trackable.h Added some doxygen documentation.
2004-07-09 Murray Cumming <murrayc@murrayc.com>
* tests/: Added test_disconnect_during_emit.cc, to prove that this
works.
2004-07-08 Murray Cumming <murrayc@murrayc.com>
* tests/test_retype_return.cc: foo::operator(int): return a
value. The SUN Forte 5.5 compiler complains about this, as it should.
2004-07-08 Murray Cumming <murrayc@murrayc.com>
* sigc++/macros/signal.h.m4: class signal*: Rename the slot_list
typedef to slot_list_type, because there is already a template class
called slot_type. SUN Forte 5.5 seems to complain about this and I am
not surprised. The old typdef is still there for backwards
compatibility, except when building with SUN Forte.
2004-07-07 Murray Cumming <murrayc@murrayc.com>
* scripts/cxx.m4: SIGC_CXX_TEMPLATE_SPECIALIZATION_OPERATOR_OVERLOAD():
Don't define the SIGC_TEMPLATE_SPECIALIZATOIN_OPERATOR_OVERLOAD C
macro at all if the test fails. This might fix the build on SUN Forte.
* sigc++/functors/macros/mem_fun.h.m4: Default constructor: Initialize
the func_ptr_ member variable. I have no evidence that this solves any
problems, but it worried me.
* sigc++/functors/slot_base.h: operator bool(): Correct documentation,
to use @code instead of <code>
* sigc++/macros/signal.h.m4: Remove the documentation for the
parameters named first and last, because they do not exist.
2004-05-31 Martin Schulze <mschulze@cvs.gnome.org>
* autogen.sh: Add '--force'-flag to the 'libtoolize'-command (bug #143425).
2.0.3:
2004-05-30 Martin Schulze <mschulze@cvs.gnome.org>
* configure.ac: Bump version number to 2.0.3.
* NEWS: Add ChangeLog summary for version 2.0.3.
* sigc++/macros/signal.h.m4: Fix segfault on emission of unconnected signal.
* tests/test_signal.cc, tests/test_accumulated.cc: Emit unconnected signal.
* sigc++/macros/object_slot.h.m4: Suppress compiler warning at
dynamic_cast<>-test (tested by Christof Petig/Timothy M. Shead).
2.0.2:
2004-05-22 Martin Schulze <mschulze@cvs.gnome.org>
* configure.ac: Bump version number to 2.0.2.
* NEWS: Add ChangeLog summary for version 2.0.2.
2004-05-20 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/macros/signal.h.m4: If a custom accumulator is specified
invoke it on signal emission even if the signal's slot list is empty.
(This used to be the case in libsigc++-1.2 as pointed out by Timothy.)
2004-05-20 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/macros/object_slot.h.m4: Suppress compiler warning at
dynamic_cast<>-test (suggested by Timothy M. Shead).
2004-05-01 Martin Schulze <mschulze@cvs.gnome.org>
* README: Updated for libsigc++-2.0.
2.0.1:
2004-04-27 Martin Schulze <mschulze@cvs.gnome.org>
* configure.ac: Bump version number to 2.0.1.
* NEWS: Add ChangeLog summary for version 2.0.1.
* sigc++/adaptors/lambda/macros/base.h.m4: Fixed documentation.
* sigc++/adaptors/macros/bind.h.m4: Hide work-arounds from doxygen.
* scripts/cxx.m4, sigc++config.h.in, configure.ac,
sigc++/adaptors/macros/bind.h.m4: Removed configure check. It
showed that the Apple gcc can also compile the sophisticated version
of the work-around.
2004-04-26 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/macros/object_slot.h.m4: Modified test for SigC::Object
inheritance so that it also works if SigC::Object is virtual base.
(Fixes bug 141094 reported by Jonathan Brandmeyer)
2004-04-26 Martin Schulze <mschulze@cvs.gnome.org>
* scripts/cxx.m4: Updated the configure check. It would probably
have succeeded on the Apple.
2004-04-26 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/adaptors/macros/bind.h.m4: Add work-arounds for
bind<-1>::deduce_result_type compilation error on Apple gcc 3.3.
* scripts/cxx.m4, sigc++config.h.in, configure.ac: Add configure
check for the compilation error above.
* sigc++/adaptors/lambda/macros/operator.h.m4: Replace _A with
_Aa. _A is a reserved keyword on Apple gcc 3.3 (Spundun Bhatt).
(fixes bug #10444 reported by Spundun Bhatt)
2004-04-19 Martin Schulze <mschulze@cvs.gnome.org>
* sigc++/signal_base.cc: Fixed serious bug in signal_base::impl():
Only reference a newly created object (initial reference).
(This fixes bug #140269 reported by Andris.)
2004-04-19 Murray Cumming <murrayc@murrayc.com>
* scripts/cxx.m4: Updated the operator() template check, because it
failed with gcc 3.4 (from cvs). Apparently the template keyword can
only be used from another template.
2.0.0:
2004-04-06 Martin Schulze <mschulze@cvs.gnome.org>
* configure.ac: Bump version number to 2.0.0.
* NEWS: Add ChangeLog summary for version 2.0.0.
* TODO, AUTHORS: Bring up to date.
* sigc++-2.0.pc.in, Makefile.am: 1.9 -> 2.0
* Added more documentation.
2004-04-10 Murray Cumming <murrayc@murrayc.com>
* sigc++/connection.[h|cc]: Implement blocked() to avoid undefined
symbol linker error.
2004-04-08 Murray Cumming <murrayc@murrayc.com>
* dist the scripts directory.
1.9.16:
2004-04-06 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.16.
* NEWS: Add ChangeLog summary for version 1.9.16.
2004-04-02 Murray Cumming <murrayc@murrayc.com>
* sigc++/connection.cc: Make block() and unblock() always return a
value, to fix compiler warnings. Patch from bug #138620 by
Alexander Nedotsukov.
2004-04-02 Murray Cumming <murrayc@murrayc.com>
* Fix the compile of examples/member_method.cc. Bug #131701 from
Kirill Smelkov. I also made the examples build as part of the regular
build.
2004-04-02 Murray Cumming <murrayc@murrayc.com>
* sigc++config.h.m4: Remove every undef apart from the one we need, to
avoid clashes, because we #include this in a public header.
2004-03-25 Murray Cumming <murrayc@murrayc.com>
* scripts/cxx.m4, configure.in, sigc++config.h.in: Rename the
template_keyword check to template_specialization, because the problem
is with or without the keyword.
* sigc++/adaptors/macros/adaptor_trait.h.m4: Define
SIGC_WORKAROUND_OPERATOR_PARENTHESES, which calls either operator() or
sun_forte_workaround() depending on the result of the compiler test.
* many .m4 files: Add sun_forte_workaround methods that call the
operator() methods. Put them in #ifdefs so that only SUN Forte C++
sees them.
2004-03-22 Murray Cumming <murrayc@murrayc.com>
* Makefile.am, sigc++/Makfile.am: Fix the sigc++config.h.in disting,
to fix make distcheck.
2004-03-21 Murray Cumming <murrayc@murrayc.com>
* Rename config.h.in to sigc++config.h.in so that gtkmm does not
include some other config.h at strange times - fixes a problem in
the gtkmm demos. This should really be in the sigc++ directory, but
that seems to add that as an include path, which causes the STL
headers to include sigc++/signal.h instead of some STL signal.h header.
2004-03-20 Murray Cumming <murrayc@murrayc.com>
* Makefile.am: Install the config.h platform-specific header.
* sigc++-2.0.pc.in: Report the include path for config.h
2004-03-20 Murray Cumming <murrayc@murrayc.com>
* Added config.h.in, using autoheader, from which config.h will be
generated, so we can detect compiler features.
* configure.ac: Added AC_CONFIG_HEADER(config.h) to generate config.h
from config.h.in.
* scripts/cxx.m4: Added this directory and file, with a
SIGC_CXX_TEMPLATE_KEYWORD_OPERATOR_OVERLOAD macro that defines
the SIGC_TEMPLATE_KEYWORD_OPERATOR_OVERLOAD C macro.
* autogen.sh: Added -I scripts to the aclocal call, so that it finds
the m4 macro for configure.ac.
* sigc++/adapators/macros/adaptor_trait.h.m4: Include config.h and
use SIGC_TEMPLATE_KEYOWRD_OPERATOR_OVERLOAD.
2004-03-18 Martin Schulze <mschulze@cvs.gnome.org>
* tests/test_mem_fun.cc, tests/test_ptr_fun.cc: Don't test
making functors from overloaded methods with partial template
specialization. Not portable among different compilers (SUN FORTE).
* adaptors/macros/apdaptor_trait.h.m4: Only gcc seems to use the
notation A.template operator()<...>(...) => adapt preprocessor check
for #define LIBSIGC_TEMPLATE_PREFIX. TODO: replace with configure check.
2004-03-13 Murray Cumming <murrayc@murrayc.com>
* g++ 3.4 (pre-release) build fixes:
* sigc++/macros/signal.h.m4: slot_iterator_buf::operator*():
Use blocked() and empty() instead of non-existant blocked_and_empty().
* sigc++/functors/macros/mem_fun.h.m4: memfun_functor*::operator()():
Use this->func_ptr_ instead of just func_ptr_.
* sigc++/adaptors/macros/deduce_result_type.h.m4: Use
T_functor::template deduce_result_type<> instead of just
T_functor::deduce_result_type<>.
* sigc++/adaptors/lambda/macros/base.h.m4, operator.h.m4, group.h.m4::
Use template keyword again. operator[](): Use this->value_ instead of
just value_.
* sigc++/adaptors/lambda/macros/bind/m4: Use template keyword, and
this-> again.
* sigc++/adaptors/macros/compose.h.m4, hide.h.m4, bind_return.h.m4,
exception_catch.h.m4:
rettype.h.m4, rettype_return.h.m4: Use template keyword,and this-> again
1.9.15:
2004-02-27 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.15.
* NEWS: Add ChangeLog summary for version 1.9.15.
2004-02-27 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functors/macros/slot.h.m4: Make the unnumbered slot templates'
copy ctors use the copy ctors of the base class. Fixes bug #24698.
* tests/test_slot.cc: Test copy ctor (Bryan Forbes).
2004-02-27 Martin Schulze <teebaum@cvs.gnome.org>
* tests/type_functor_trait.cc: Bring it up-to-date (use sigc::ref).
Make it work with the SUN Forte.
2004-02-24 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/type_traits.h: Make is_base_and_derived<> work with the SUN Forte.
2004-02-19 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/type_traits.h: Make is_base_and_derived<> platform independant.
* sigc++/adaptors/lambda/macros/base.h.m4: Make lambda_core<> ctors
explicit. Remove an unused ctor from lambda_core<T_type, true>.
2004-02-14 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functors/slot_base.h, sigc++/functors/macros/slot.h.m4:
Move some documentation to slot_base.h.
* sigc++/signal_base.h, sigc++/macros/signal.h.m4:
Move some documentation to signal_base.h.
API addition:
* sigc++/functors/macros/slot.h.m4: Add numbered slot# templates.
Make unnumbered slot templates inherit from slot#.
API change:
* sigc++/functors/macros/mem_fun.h.m4: Allow for methods of the object's
base types to be passed into sigc::mem_fun(). (Used to be the case in
libsigc++-1.2).
2004-02-13 Murray Cumming <murrayc@usa.net>
* sigc++/functors/slot_base.[h|cc], sigc++/trackable.[h|cc]: Create
and use a typedef for the destroy_notify callback functions, to avoid
confusion function pointer declaration syntax in the API.
2004-02-13 Murray Cumming <murrayc@murrayc.com>
* Moved implementation to .cc files:
* sigc++/functors/: Added slot_base.[h|cc] which contains non-template
code that was previsouly in the generated functors/slot.h and
non-generated slot.cc files. All non-inline implementation is now in
the .cc file.
* sigc++/functors/macros/slot.m4: Removed the code that has been moved
to slot_base.[h|cc].
* sigc++/: Added signal_base.[h|cc] which contains non-template code
that was previously in the generated signal.h and non-generated
signal.cc file. All non-inline implementation is now in the .cc file.
* sigc++/macros/signal.m4: Removed the code that ahs been moved to
signal.cc
* sigc++/connector.[h|cc]: method implementation moved to the .cc file.
1.9.14:
2004-02-13 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.14.
* NEWS: Add ChangeLog summary for version 1.9.14.
2004-02-09 Murray Cumming <murrayc@usa.net>
* sigc++/functors/macros/slot.h.m4: slot_base: Added operator bool(), needed to
check for a slot that is created with the default constructor. This was
present in libsigc++ 1.2 also.
2004-02-06 Murray Cumming <murrayc@usa.net>
* Makefile.am: Build the docs directory, by adding it to SUBDIRS.
* docs/Doxyfile.in: Updated to be more glibmm-like.
* Added some @deprecated doxygen bits.
* sigc++/macros/signal.h.m4: Call base constructor from signal_base
constructor - this is an error-as-warning when building gtkmm.
1.9.13:
2003-11-30 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.13.
* NEWS: Add ChangeLog summary for version 1.9.13.
* Makefile.am, MSVC_Net2003/Makefile.am, configure.ac:
Distribute MS .Net project files.
* sigc++/adaptors/macros/[bind,hide].h.m4: Correct and add
documentation. Make hide_functor ctor explicit.
2003-11-11 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors/macros/[bind,hide].h.m4: Change to zero-based
argument index in numbered bind() and hide() overload
(Agreement on the mailing list).
Support binding up to CALL_SIZE arguments with one bind adaptor.
(Requested by joey yandle and others).
Only support binding of multiple arguments in unnumberd overloads
to keep the API simple (Requested by Murray Cumming).
* tests/test_[bind,hide,functor_trait].cc, sigc++/bind.h:
Reflect API changes in compatibility module and test cases.
2003-11-10 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors/macros/[bind,hide].h.m4: Add unnumbered
bind() and hide() overloads to make specifying the argument
position optional (Proposed by Jeff Franks).
* tests/test_[bind,hide].cc: Test unnumbered bind() and hide().
* sigc++/adaptors/macros/adaptor_trait.h.m4:
Change "#ifdef MSVC" to "#ifdef _MSC_VER" (Roel Vanhout).
2003-11-09 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functors/macros/slot.h.m4: Change functor type in
typed_slot_rep to adaptor_trait::adaptor_type<T_functor>
and use explicit function template instantiation in
internal::slot_call::call_it(). Avoids copying of arguments
in call_it() and enables binding of object instances
to class methods through bind() (Reported by Jeff Franks).
* tests/test_bind.cc: Test binding object instances to
class methods through bind().
* sigc++/adaptors/adaptors.h: Include retype[_result].h.
* sigc++/adaptors/macros/adaptor_trait.h.m4:
- Add documentation.
- Mark some c-tors as explicit.
- Remove ununsed operator T_functor&() from adaptor_functor.
* sigc++/adaptors/macros/deduce_result_type.h.m4:
Rewrite parts of the documentation.
* sigc++/adaptors/macros/bind.h.m4: Add documentation.
* sigc++/functors/macros/mem_fun.h.m4: Remove unnecessary
explicit markers. Minor fixes to documentation.
* sigc++/functors/macros/functor_trait.h.m4:
Minor fixes to documentation.
1.9.12:
2003-11-04 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.12.
* NEWS: Add ChangeLog summary for version 1.9.12.
2003-11-03 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/macros/signal.h.m4: Document accumulators.
Move slot_iterator_buf into namespace internal. Since
accumulators have the iterator type as a template argument
there is no need to expose this very internal type.
* sigc++/functors/macros/*.m4: Regroup documentation.
Documentation of the core parts of the library should be
complete by now.
2003-11-02 Martin Schulze <teebaum@cvs.gnome.org>
* Improve documentation of the core parts of the library.
* tests/test_ptr_fun.cc: Test ptr_fun() with static
member functions.
2003-11-02 Martin Schulze <teebaum@cvs.gnome.org>
* Move all .m4 files into new subdirectories
sigc++/[...]/macros. Install the .m4 files on
"make install" (Reported by Ron Steinke).
2003-11-01 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/[class,method,object]_slot.h.m4: Include
sigc++/functors/mem_fun.h (Reported by Ron Steinke).
2003-11-01 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors/lambda/operator.h.m4: Add negation
operator I have completely overlooked until now.
* sigc++/tests/test_lambda.cc: Test negation operator.
2003-11-01 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/[class_slot,method_slot,object_slot,hide].h.m4,
sigc++/signal.h.m4, sigc++/functors/mem_fun.h.m4:
- Use a shorter notation for ..._mem_function variants.
- Change order of mem_funcotr<>'s template arguments to
match std::mem_fun_t and to be more consistent with adaptors.
- Use ::sigc::slot's implicit ctor in compatibility module.
* sigc++/adaptors/lambda/operator.h.m4: Change some
lambda action names to match action names in std.
API addition:
* sigc++/adaptors/retype.h.m4: New file adding
adaptor retype.
* sigc++/Makefile.am: Build and distribute new file.
* tests/test_retype.cc: New file testing adaptor retype.
* MSVC_Net2003/tests/test_retype/test_reytype.vcproj,
tests/Makefile.am: Build and distribute new test case.
2003-11-01 Martin Schulze <teebaum@cvs.gnome.org>
* MSVC_Net2003: New directory containing project
files for Visual Studio .Net 2003.
Credits to Roel Vanhout <roel@riks.nl>!
2003-11-01 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/retype.h.m4: Use LIBSIGC_TEMPLATE_PREFIX
in explicit function template instantiations.
* sigc++/type_traits.h: Add template specialization
for arrays (T_type[N]) to disable non-working member
type_trait<T_type[N]>::instance().
* sigc++/visit_each.h: Remove more disturbing
limit_derived_target<>::operator() overloads.
(Should have noticed earlier that they are unnecessary.)
* sigc++/adaptors/deduce_result_type.h.m4,
sigc++/adaptors/lambda/operator.h.m4,
sigc++/functors/functor_trait.h.m4,
tests/test_[bind,compose,exception_catch,hide,lambda].cc:
Completely removed support for typeof(). We don't need
it any more and it is completely non-standard!
2003-10-30 Cedric Gustin <cedric.gustin@swing.be>
* configure.ac: Added test of win32 platform. Commented
out AM_DISABLE_SHARED (DLLs are shared libraries).
* sigc++/Makefile.am: added a few LDFLAGS for win32
DLLs.
2003-10-30 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/signal.h.m4: Add SigC::Signal#<>::slot().
* sigc++/slot.h.m4: Comment out make_slot() work-around.
* sigc++/adaptors/bind.h.m4: Remove unnecessary brackets
in template argument lists. They are confusing MSVC.
* sigc++/adaptors/*.h.m4, sigc++/adaptors/lambda/*.h.m4:
Use LIBSIGC_TEMPLATE_PREFIX in explicit function
template instantiations.
* sigc++/tests/test_*.cc:
- Include <string> where std::string is used.
- Use double instead of float.
2003-10-27 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/retype.h.m4: Cleanup.
* TODO: Bring it up to date.
1.9.11:
2003-10-26 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.11.
* NEWS: Add ChangeLog summary for version 1.9.11.
2003-10-26 Martin Schulze <teebaum@cvs.gnome.org>
Compatiblity module:
* sigc++/signal.h.m4: Move definition of compatiblity
classes SigC::SignalN to here.
* sigc++/connection.h:
- Add connection::connected().
- Add compatibility typedef SigC::Connection.
* sigc++/bind.h, sigc++/bind_return.h,
sigc++/class_slot.h.m4, sigc++/hide.h.m4,
sigc++/method_slot.h.m4, sigc++/object.h,
sigc++/object_slot.h.m4, sigc++/retype.h.m4,
sigc++/retype_return.h sigc++/slot.h.m4,
sigc++/compatibility.h:
New files to complete compatibility module.
Split content of compatibility.h.m4 among the new files.
* sigc++/compatibility.h.m4: Removed.
* Makefile.am: Build and distribute new files.
* tests/test_compatibility.cc: Test new stuff.
Fixes:
* sigc++/functors/slot.h.m4: Fix copy constructor and
operator=() of slot template.
* sigc++/adaptors/bind.h.m4: Fix deduce_result_type
template specializations. bind<0>() probably compiles
with gcc-3.3, now.
2003-10-26 Martin Schulze <teebaum@cvs.gnome.org>
Fixes:
* sigc++/functors/slot.{cc,h.m4}:
- Fix notification process: don't defer detaching of a
slot from all referred trackables during signal emission!
- Size optimization: replace virtual functions from
struct typed_slot_rep with function pointers in slot_rep
(reduces size of a typical typed_slot_rep instantiation
by 30% !!!).
* tests/test_slot.cc: Test sigc::slot more thoroughly.
* sigc++/functors/mem_fun.h.m4: Fix visit_each().
* sigc++/adaptos/bind_return.h.m4: Add support for
sigc::ref().
* tests/test_bind_return.cc: Use sigc::ref().
* sigc++/signal.h.m4: Avoid compiler warning about
uninitialized variable r_ in emit().
* sigc++/visit_each.h: Cleanup.
API additions:
* sigc++/adpators/lambda/operators.h.m4: Add
lambda actions sigc::{reinterpret,static,dynamic}_cast_
to support explicit parameter conversion.
* tests/test_lambda.cc: Test sigc::static_cast_.
* sigc++/adaptors/retype_return.h.m4: New file adding
adaptor retype_return (and hide_return).
* sigc++/Makefile.am: Build and distribute new file.
* tests/test_retype_return.cc: New file testing
adaptor retype_return (and hide_return).
* tests/Makefile.am: Build and distribute new test case.
2003-10-25 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/visit_each.h: Remove disturbing
limit_derived_target<>::operator() overloads.
* sigc++/adaptors/bind.h.m4: Add support for sigc::ref().
* tests/test_bind.cc: Test sigc::ref().
* sigc++/adaptors/lambda/{operator,group,base}.h.m4:
- Move support for sigc::ref() from lambda_core<> into
lambda operator and lambda group creator functions.
- Add missing visit_each() overload for lambda<> template.
* tests/test_lambda.cc: Test auto-disconnection.
TODO: Fix a strange bug that leads to "Bus error"
during auto-disconnection.
1.9.10:
2003-10-23 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.10.
* NEWS: Add ChangeLog summary for version 1.9.10.
2003-10-23 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functors/{functor_trait,slot}.h.m4:
Move definition of struct nil into functor_trait.h.
2003-10-23 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Disable typeof() compiler checks.
* sigc++/adaptors/bind.h.m4: Remove unnecessary
deduce_result_type<> template specializations.
2003-10-20 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors/compose.h.m4:
Correct order of typedefs for good. (Patch from Jeff Franks.)
1.9.9:
2003-10-20 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/connection.h: Add constructor that takes
a sigc::slot_base& to support 3rd-party slot lists
like they are used in glibmm/gtkmm.
* sigc++/functors/slot.h.m4: Make sigc::slot::call_type public.
(Fixes compile problems reported by Jeff Franks.)
* sig++/type_traits.h: Don't use long long in
sigc::is_base_and_derived.
(Fixes compile problems reported by Jeff Franks.)
* sigc++/adaptors/{bind,compose,hide,exception_catch}.h.m4:
Correct order of typedefs. (Repoted by Jeff Franks.)
* configure.ac: Bump version number to 1.9.9.
* NEWS: Add ChangeLog summary for version 1.9.9.
1.9.8:
2003-10-19 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functors/slot.h.m4: Define doxygen group functors.
* configure.ac: Bump version number to 1.9.8.
* NEWS: Add ChangeLog summary for version 1.9.8.
2003-10-19 Martin Schulze <teebaum@cvs.gnome.org>
* NEWS: Add announces of versions 1.9.6 and 1.9.7.
* sigc++/compatibility.h.m4: New file. Defines namespace SigC.
namespace SigC should be API compatible to libsigc++-1.2.
* sigc++/Makefile.am: Build compatibility.h.
* tests/test_compatibility.cc, tests/Makefile.am:
Add test case for compatibility module.
* docs/index.html: Change group names.
* sigc++/sigc++.h: Include connection.h.
* sigc++/connection.{cc,h}:
- Rename dependency to destroy_notify_callback.
- Change parameter name in set_slot() from d to data.
- Fix operator=(): Add "return *this;"
- Get rid of namespace functor.
- Corrections in documentation.
* sigc++/signal.{cc,h.m4}:
- Add reference counter to signal_impl. Replaces "bool destroy_".
- Move signal_base, slot_iterator[_buf], slot_list out of
namespace internal. They are part of the public API.
- Add convenience function signal#::make_slot().
- Get rid of namespace functor.
- Corrections in documentation.
* sigc++/trackable.{cc,h}:
- Rename dependency to destroy_notify_callback.
- Rename trackable::clear() to trackable::notify_callbacks().
- Corrections in documentation.
* sigc++/type_traits.h: Add documentation.
* sigc++/visit_each.h:
- Get rid of namespace functor.
- Add documentation.
* sigc++/adaptors[/lambda]/*: Get rid of namespace functor.
* sigc++/functors/{functor_trait.h,ptr_fun.h.m4,mem_fun.h.m4}:
- Get rid of namespace functor.
- Corrections in documentation / add documentation.
* sigc++/functors/slot.{cc,h.m4}:
- Move slot_base out of namespace internal. It's public API.
- Get rid of one-letter-parameter-names.
- Get rid of namespace functor.
- Corrections in documentation.
* tests/*.cc: Get rid of "using namespace ...".
2003-09-10 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors/lambda/{base,operators}.h.m4:
Add subscript ([]) and assign (=) operator. I think there are now
enough operators available to make any future power user happy.
The only one missing is the comma operator and if we added it
the logical consequence would be to also add if(), switch(), do(),
etc. lambda expressions which are really out of place in libsigc++.
* sigc++/type_traits.h: Fix is_base_and_derived<> for const types.
* tests/test_lambda.cc: Test new operators.
1.9.7:
2003-09-05 Martin Schulze <teebaum@cvs.gnome.org>
* configure.ac: Bump version number to 1.9.7.
2003-09-03 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors/lambda/operator.h.m4:
- Restructure so that the size of the generated source file stays
reasonable for 34 operators: There are only two lambda templates
lambda_operator and lambda_operator_unary. The action is an additional
template parameter. A small template lambda_action[_unary] is specialized
for all actions.
- Add most operators that boost::lambda supports. Missing operators are
"=", "[]", "," and support for pointer arithmetic. I don't know if it's
worth adding these. In libsigc++, the purpose of lambda operators is to
provide some extra functionality for the group adaptor.
* tests/test_lambda.cc:
Test pre-increment, address and dereference operator.
2003-08-31 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/reference_wrapper.h, sigc++/type_traits.h, sigc++/Makefile.am:
New file reference_wrapper.h provides ref() to specify that adaptors/lambdas
should take a reference to the object passed into ref() instead of a copy.
* tests/test_lambda.cc:
- Test group() with mem_fun().
- Use ref() where lambdas should store references to objects.
- Test var() and constant().
* sigc++/adaptors/lambda/base.h.m4:
- Support ref() in return type deduction.
- Add var() and constant() which create lambdas for usage with lambda operators.
* sigc++/adaptors/lambda/operator.h.m4:
- Fix return type deduction.
- Remove operator{+,-,*,...} overloads added on 2003-08-29. ref() is way cleaner.
* sigc++/adaptors/lambda/group.h.m4,
sigc++/adaptors/bind.h.m4, sigc++/adaptors/compose.h.m4,
sigc++/adaptors/exception_catch.h.m4, sigc++/adaptors/hide.h.m4:
Fix return type deduction.
2003-08-29 Martin Schulze <teebaum@cvs.gnome.org>
* tests/test_lambda.cc: Add more tests.
* sigc++/adaptors/lambda/select.h.m4, sigc++/adaptors/lambda/lambda.cc.m4:
Make _1, _2, ... constant.
* sigc++/adaptors/lambda/operator.h.m4:
Add operator{+,-,*,...} overloads to distinguish between const and non-const objects.
Store references to non-const objects rather than copies.
This allows expressions like e.g. std::cout << _1.
* sigc++/adaptors/lambda/base.h.m4, sigc++/adaptors/lambda/group.h.m4:
Remove void specializations. Functors returning void are tested and work fine.
2003-08-27 Martin Schulze <teebaum@cvs.gnome.org>
* tests/test_callof.cc, tests/test_deduce_result_type.cc:
Rename, correct and improve this test case.
* tests/Makefile.am: Build and run test_deduce_result_type
instead of test_callof.
2003-08-27 Martin Schulze <teebaum@cvs.gnome.org>
* Update TODO.
2003-08-27 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors/hide.h.m4: Remove usage of callof_ignore_arg<>.
* sigc++/callof.h.m4, sigc++/adaptors/deduce_result_type.h.m4,
sigc++/functors/functor_trait.h.m4:
Remove the first and move deduce_result_type templates from
functor_trait.h.m4 into new file deduce_result_type.h.m4.
* sigc++/Makefile.am, sigc++/sigc++.h, sigc++/adaptors/adaptor_trait.h.m4:
Build and include sigc++/adaptors/deduce_result_type.h instead of callof.h.
* sigc++/functors/slot.h.m4: Document struct nil.
2003-08-24 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functors/functor_trait.h.m4: Simplify usage of convenience
macro SIGC_FUNCTORS_HAVE_RESULT_TYPE:
namespace sigc{ namespace functor{ SIGC_FUNCTORS_HAVE_RESULT_TYPE }}
2003-08-24 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functors/functor_trait.h,m4, sigc++/adaptors[/lambda]/*.h.m4:
Merge adaptor return type deduction and typeof() into
sigc::functor::deduce_result_type. Use it for all adaptors.
* tests/test_compose.cc: Only test multi-type get-functor if
typeof() if supported.
2003-08-24 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/adaptors[/lambda]/*.h.m4:
- Remove unnecessary void specializations. In fact, only the one
for sigc::functor::exception_catch<> is needed and I don't really
understand why. For the lambda stuff the void specializatoins are
just commented out at the moment.
- Make typeof() optional. Surprisingly, I got the lambda stuff working
without typeof()! The test suite doesn't catch all cases yet, so maybe
some thing are still not working.
TODO: Implement configure check.
* tests/test_bind.cc, tests/test_compose.cc tests/test_exception_catch.cc,
tests/test_hide.cc, tests/test_lambda.cc:
Only test multiple functor return types if typeof() is supported.
2003-08-06 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/trackable.{cc,h}: Add function trackable::clear().
2003-06-24 Andreas Rottmann <rottmann@users.sourceforge.net>
* TODO: Minor tweaks.
2003-06-23 Andreas Rottmann <rottmann@users.sourceforge.net>
* docs/reference/Doxyfile.in: Use these variables.
* docs/reference/Makefile.am (html/index.html): Provide doxygen
with SRCDIR and TOP_SRCDIR environment variables.
* sigc++/functors/slot.h.m4: Make slot::call_type typedef public;
this fixes a g++ 3.3 error in signal.h.
* sigc++/signal.h.m4: Make the signal::accumulated class public;
this fixes a g++ 3.3 error in test_accumulated.cc.
2003-06-15 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/functor/slot.h.m4: Correct typing error in docs.
* sigc++/functor/ptr_fun.h.m4: Document the whole thing.
2003-05-31 Murray Cumming <murrayc@usa.net>
* Reference documentation: Rearranged the groups to make it all
a bit more like the libsigc++ 1.2 reference documentation.
Corrected some spelling and grammar too.
This needs a lot of work. The text is very hard to read and it's
generally not appropriate for a user of the code who doesn't
care about the internals. But it's not impossible - our examples
should show us what we need to say in the documentation.
We probably need some more groups for the extra stuff, like we do
in libsigc++ 1.2.
2003-05-29 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/signal.h.m4: Fix documentation.
* sigc++/connection.h, sigc++/functor/slot.h.m4:
Document the whole thing.
2003-05-29 Martin Schulze <teebaum@cvs.gnome.org>
* sigc++/signal.h.m4:
- Remove bogus operator() from unnumbered signal<> and
signal<>::accumulated templates.
- Document the whole thing.
* docs/index.html: Fix some links.
2003-04-06 Martin Schulze <MHL.Schulze@t-online.de>
* TODO, configure.ac, Makefile.am, docs/*:
Add Doxygen framework.
2003-04-06 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/callof.h.m4, sigc++/adaptors/*, tests/test_callof.cc:
Move sigc::callof<> to sigc::functor::internal::callof<>.
* sigc++/functors/mem_fun.h.m4, tests/test_mem_fun.cc:
Add new types [bound_][const_]volatile_mem_functor, visit_each()
and mem_fun() overloads for volatile qualifier.
Add ctor overloads in bound_*mem_functor and mem_fun() overloads
that take reference instead of pointer.
2003-03-26 Martin Schulze <MHL.Schulze@t-online.de>
* Change "closure" to "slot" throughout sigc++2 (file names,
class names, member variables, documentation, etc.).
2003-03-26 Martin Schulze <MHL.Schulze@t-online.de>
* TODO: Rewrite to reflect recent changes as well as recent discussions.
2003-03-24 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/adaptors/bind_return.h.m4: Make the adaptor's data member
public so that visit_each() can access it.
* sigc++/adaptors/lambda/*.h.m4: More fixes. Add a note about
malfunctioning typeof() (probably compiler bug in gcc-3.2).
* tests/*.cc: Test references. Fix compose equivalent in test_lambda.
2003-03-24 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/Makefile.am, sigc++/functors/functor_trait.h[.m4],
sigc++/adaptors/adaptor_trait.h.m4: Move detection of function
and member method pointers' return types from adaptor_trait into
functor_trait. (We'll use functor_trait rather than adaptor_trait for
our lambda stuff.) functor_trait.h needs to be generated from .m4 now.
* sigc++/functors/functor_trait.h.m4: Add convenience macros:
- SIGC_FUNCTORS_HAVE_RESULT_TYPE indicates that the existance of
T_functor::result_type should be assumed for all unknown functors.
- SIGC_FUNCTOR_TRAIT(T_functor, T_result) explicitly specifies the
result type of a functor.
("typename functor_trait<T_functor>::result_type") is used to
determine the return type of our adaptors' operator()() overloads.
* sigc++/adaptors/[lambda/]*.h.m4: Various fixes in visit_each() and
operator()() overloads to make these operator()() overloads usable.
Most of them were just commented out before. Some adaptor types also
have void specializations, now.
* sigc++/adaptors/lambda/group.h.m4: Change syntax from
"[some_functor] % grp([args])" to "group([some_functor], [args])"
like we agreed on the ml some time ago.
* sigc++/tests/test_[all adaptors].cc: Test stuff that didn't work
before.
2003-03-22 Murray Cumming <murrayc@usa.net>
* Added pgk-config file, from a mystery person in bug #108857
2003-03-22 Martin Schulze <MHL.Schulze@t-online.de>
* tests/test_bind.cc: Test and show how to use functor_trait
for user defined or 3rd-party functors so that a
bind<0>([functor],[arg1])() call with no arguments can return a value.
2003-03-20 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/callof.h.m4: Add explanations. Comment in / create templates
callof_safe[#]. Unfortunately they don't work for functors with overloaded
operator() so we don't use it for now. At least everything is documented.
* sigc++/functors/functor_trait.h, sigc++/functors/*.h.m4: Add back
functor_base compiler hint. We're using it now in adaptor_functor<>.
* sigc++/adaptors/{adaptor_trait,bind}.h.m4: Make operator() overloads
with no arguments return the result of the functor invocation.
Fix multiple bind<0>().
* tests/test_bind.cc: Test the stuff that's working now.
2003-03-16 Murray Cumming <murrayc@usa.net>
* Added sigc++/sigc++.h, like in libsigc++ 1.2
* examples: Added member_method example, which uses a class method
and which demonstrates disconnection.
1.9.6:
2003-03-11 Andreas Rottmann <rottmann@users.sourceforge.net>
* sigc++/Makefile.am: Use substitution references instead of
$(patsubst). Is shorter and fixes the strange-dirs-in-dist-tarball
bug.
2003-03-09 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/connection.h: Add block() capability.
2003-03-09 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/signal.{cc,h.m4}: Add flag signal_impl::destroy_
and function signal_impl::destroy(). Use them to defer
the destruction of the signal_impl object during signal
emission.
* tests/test_disconnect.cc: Add tests for the connection
class and for deleting signals during emission.
2003-03-09 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/connection.{cc,h}, sigc++/Makefile.am:
- New files that add a connection class. Objects of this
class are constructed from closure list iterators and can
be used to disconnect the refered closure. As opposed to
iterators they stay valid beyond the lifetime of the closure.
2003-03-09 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/functors/closure.{cc,h.m4}, sigc++/signal.cc:
- Rename closure_{base,rep}::[set_]dependency_ -> [set_]parent_.
- Make closure_rep inherit trackable. This allows for
connection objects that stay valid beyond the life time
of the refered closure.
- Make some one-line-functions inline again.
2003-03-08 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/trackable.cc: BUGFIX in trackable_dep_list::clear()
2003-03-08 Andreas Rottmann <rottmann@users.sourceforge.net>
* sigc++/Makefile.am: Rewritten so we can build lambda cleanly.
* sigc++/Makefile.am_fragment: Removed.
* sigc++/functors/Makfile.am: Removed.
* sigc++/adaptors/Makefile.am: Removed.
* sigc++/adaptors/lambda/Makefile.am: Removed.
* configure.ac (AC_OUTPUT): Remove the above Makefiles.
* tests/Makefile.am: Re-included lambda test.
2003-03-07 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/signal.{cc,h.m4}:
- signal_emit#<>: New templates replacing signal#<>::caller.
The purpose of these templates is implementing the emit
function and optimizing signal emission for the case that
no accumulator is used via template specializations.
- default_accumulator<>: Removed. The default for T_accumulator
in signal#<> now is nil. An example how to use accumulators
is available in tests/test_accumulator.cc.
- signal_{base,impl}: Move the implementation of signal_base's
interface to signal_impl. An object of this class is
dynamically allocated when first connecting a closure to
the signal. This drastically reduces the size of empty signals
and allows for future addition of a reference counter to make
it safe to delete a signal during emission.
- Directly work on closure_rep during signal emission. This
seems to be quicker than using the closure templates.
- Document the classes. Restructure the header file a bit.
* sigc++/functors/closure.h.m4: Make closure_base::rep_ data
member public, so that signal emission can directly work on it.
* tests/test_size.cc: Add an entry for signal_impl.
2003-03-07 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/functors/closure.{cc,h.m4}:
- closure_base: BUGFIXES in ~closure_base() and operator=().
- Mark some functions with the inline keyword. This seems to
help gcc 3.2 to optimize signal emission and (dis)connection.
- Document the classes. Restructure the header file a bit.
2003-03-07 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/trackable.{cc,h}: Make trackable allocate a
trackable_dep_list object dynamically when adding the first
dependency. (This reduces the size of objects, that are not
refered by a closure by 4 bytes (50%) and increases the size
of objects that are refered by a closure by 4 bytes (50%)
on a 32 bit architecture => This reduces total memory use
when >50% of the trackables are not refered by a closure.)
Document the classes.
2003-03-05 Martin Schulze <MHL.Schulze@t-online.de>
* tests/Makefile.am, tests/test_size.cc, tests/test_accumulated.cc:
Add two test cases. test_size is showing the size of public and
internal structures. (Which apart from empty signals are smaller
than the sizes of the equivalent libsigc++-1.2 structures.)
test_accumulated is a test for the template signal<>::accumulated<>
at the same time showing the use of accumulators in libsigc++2.
* Offtopic: My note about binary sizes from 2003-02-10 is wrong.
Stripped libsigc++2 test binaries are about 8-10k in size.
2003-03-05 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/visit_each.h: BUGFIX in template specialization
limit_derive_target<T_Target*,T_action>::with_type<false,T_type>:
Add non-const overloads for static void execute_() avoiding
compile time errors.
2003-02-16 Martin Schulze <MHL.Schulze@t-online.de>
* tests/Makefile.am, tests/test_disconnect.cc:
Add testcase with a mixed connection & disconnection sequence.
2003-02-16 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/signal.cc: Bugfix in signal_base::insert():
Set notification function in the newly created copy of
slot_ rather than in slot_ itself.
2003-02-10 Martin Schulze <MHL.Schulze@t-online.de>
* sigc++/signal.h.m4: Comment in typedefs for iterator types in
the signal#<> class template. Make signal#<>::connect() return an
iterator for convenience.
(Note that the first change increases the binary size of
tests/test_signal from 201k to 204k, the second change to 206k.)
2003-01-23 Murray Cumming <murrayc@usa.net>
* sigc++/adaptors/lambda is disable temporarily (not built and
not distributed) because it gets built before its parent
directory, but #includes generated sources in the parent directory.
2003-01-22 Murray Cumming <murrayc@usa.net>
* Added Andreas Rottman's example.
2003-01-22 Murray Cumming <murrayc@usa.net>
* Applied Andreas Rottman's make dist fixes.
2003-01-14 Murray Cumming <murrayc@usa.net>
* Added whitespace to make the code more readable.