Git Log

Commit Date  
[0d8e0b] (0.4.2) by Martin Petříček Martin Petříček

bumped version to 0.4.2

2008-12-15 12:41:22 Tree
[a2f714] by Martin Petříček Martin Petříček

Commited patch by Michal to fix problem with doc/Makefile.rules missing
in tarball

2008-12-03 00:16:03 Tree
[97eb58] by Martin Petříček Martin Petříček

bumped version to 0.4.2-pre1

2008-12-03 00:14:35 Tree
[0350f2] by Michal Hocko Michal Hocko

Purge the last boost-iostreams zombies in the code.

* config/boost_iostreams.m4
- removed as it is not used anymore
- comments update
* CStream
- filtering_ostreambuf removed as it is not used anywhere
* kernel/static.h
- boost/iostreams/* includes removed and boost/noncopyable.hpp
added for using noncopyable defined later in this file
* utils/types/basic_types.h
- noncopyable removed as it is defined in boost

2008-11-25 15:11:10 Tree
[9fc4b0] by Michal Hocko Michal Hocko

*** empty log message ***

2008-11-13 14:46:06 Tree
[22134f] by Michal Hocko Michal Hocko

*** empty log message ***

2008-11-13 14:13:12 Tree
[d483e3] by Michal Hocko Michal Hocko

This patch fixes bug 262. Original implementation doesn't remove empty
directories (created during make install) during make uninstall.
The fix is based on rmdir's -p parameter which removes all empty
directory in the way up.

rmdir --ignore-fail-on-non-empty parameter may not be available on each
platform so that we don't use this parameter and rather use a new
tools/ script which emulates this behavior.

- default value for DEL_DIR updated
- set to tools/ which emulates
rmdir -p --ignore-fail-on-non-empty
* doc/Makefile
- doc_dist_uninstall updated to correctly remove DOC_PATH and
* Makefile
- uninstall-core-dev shouldn't fail when INCLUDE_PATH and LIB_PATH
don't exist
* tools/ added
- as alreadt mentioned emulates GNU rmdir -p --ignore-fail-on-non-empty
(removes all empty directories in the way up through the given
path, stop with the first non-empty directory and doesn't complain
when the directory is empty. It complains only if given directory
doesn't exist at all)

2008-11-13 14:11:39 Tree
[53a98c] by Michal Hocko Michal Hocko

Subject: [PATCH @num@/@total@] fake_install_tools introduction

This patch adds tools/ script which aimed to replace
all tools which are used during {un}installation process. Its only purpose
is to fake operation given as the first argument and log this operation with
all parameters to the file.

This can be very handy when we want to check install targets and want to
make sure that nothing is broken (I was hit several times during my

How it works?
To enable these fake tools, we just need to give --with-fake-install=YES
parameter to the configure and that's it. It will inject all variables
holding tools for installation with $(ROOT)/tools/
script and nothing more has to be done.

- --with-fake-install parameter added (NO by default)
* tools/
- script added

2008-11-13 14:09:37 Tree
[883855] by Michal Hocko Michal Hocko

Subject: [PATCH @num@/@total@] Configurable variables consolidation

It is rather long time since I have started Unified makefiles patchset and
there were several things I was lazy work on those days. Most of them were
marked as "TODO from autoconf" and the main motivation was to make almost
everything configurable from configure script rather than having them

This patch sets all those variable (in from configure.
In the result it means that we can modify them during configuration when
necessary (e.g. ./configure COPY="something else than cp" will use given
"something else than cp" instead of default "cp -f".

The second important thing which is done here, is the consolidation of all
these variables with those coming out of qmake when it generates makefile.
Qmake has its own idea about COPY, MKDIR, ... variables and we have to force
it to use our configured values.

I have used the same trick which is already used for unification of
compiler - make all configurable variables used by qmake 2 layered:
CONFIG_NAME is set by configured and used in pro file and NAME used in our
makefiles is initialized from CONFIG_NAME.

* config/env.m4
- PARAM1 stands for variable we want to check and set
- PARAM2 default value if PARAM1 isn't defined yet
- it should be used for all configurable variables because
we can push such a variable by ./configure NAME=value or
from configure itself when needed (just define such
-TODO - maybe we could include also AC_SUBST into this
macro. Any idea how Martin? (I have tried AC_SUBS($[]$1 but
it didn't work)
- all hard-coded variables from defined by CHK_OR_DEFINE_VAR
- all hard-coded variables are set from configure now (except for
C*FLAGS and our headers and libraries paths
* gui/, kpdf-kde-3.3.2/
- All variables which might affect installation replaced initialized
from CONFIG_* equivalents
- TODO I wasn't able to force SYMLINK initialization
QMAKE_SYMLINK = $(CONFIG_SYMLINK) doesn't seem to work - I have tried
even SYMLINK = $(CONFIG_SYMLINK) withou success
- information about compilation flags added

Changelog v1 -> v2
* ARCH, C{XX}_FLAGS, EXTRA_*C{XX}FLAGS set from configurable by configure
parameters in the same way like all others - hopefully no more hardcoded
variables in the
* README updated to contain information about compilation flags

2008-11-13 14:09:21 Tree
[4d6476] by Michal Hocko Michal Hocko

Subject: [PATCH @num@/@total@] Output configure files change

This patch simply changes configure-generated files in the root source
directory so they are more logical:
- -> Makefile is no longer needed, because only
[UN]INSTALL_TARGET are substitued. IMO this is not worth file
generation and moreover all install targets for doc are placed in so it is more consistent to define these
variables there
- -> Makefile.rules was added for make related
configurable stuff (like MAKEFLAGS). This way, we have
Makefile.flags for project stuff global configuration
Makefile.rules for build stuff global configuration

* removed and replaced by Makefile
- @[UN]INSTALL_TARGET@ replaced by variables defined in
- distclean updated
* Makefile.rules removed and replaced by
- configurable MAKEFLAGS added
- MAKEFLAGS removed from here because it caused qmake complains
which doesn't recognise ifeq() ... endif sequence. It is also more
logical to have it in the rules file which controls building
process rather than project related configuration
* dist-exclude
- generated files are excluded from installation

2008-11-13 14:08:50 Tree
[623ddc] by Michal Hocko Michal Hocko

*** empty log message ***

2008-11-12 12:30:10 Tree
[782edb] by Michal Hocko Michal Hocko

Bug fix for 272 - resources deallocation in error path.
Another problem coming from Delinarizator inheritance change from XRef to
CXref. We can safely remove pdfedit_core_dev_init_check and getErrorCode
checks, because they are done already in the CXref.

Delinarizator has to cleanup file handle if constructor fails, because
file handle is Delinearizator specific.
If an exception is thrown from CXref constructor, we have to cleanup also
stream because ~CXref is not called in such a path.

2008-11-12 12:29:04 Tree
[6c1837] by Jozef Mišutka Jozef Mišutka

-- finally managed to run kernel tests normally in debug mode with zlib so describing it
---- the interesting part is that it MUST be linked dynamically not statically

2008-11-11 10:50:00 Tree
[7e9f15] by Jozef Mišutka Jozef Mišutka

-- resource leak fixed

2008-11-10 14:27:51 Tree
[2eddf6] by Jozef Mišutka Jozef Mišutka

-- fix memory leak

2008-11-10 11:19:46 Tree
[8ee274] by Michal Hocko Michal Hocko

BOOST_IOSTREAMS_LIBS removed because it is no more used and configure
doesn't export BOOST_IOSTREAMS_LIB

2008-11-10 10:43:20 Tree
[0951a2] by Jozef Mišutka Jozef Mišutka

-- fix memory leak

2008-11-10 09:52:48 Tree
[69af9b] by Jozef Mišutka Jozef Mišutka

-- fix merge

2008-11-10 09:42:25 Tree
[3c9ab5] by Jozef Mišutka Jozef Mišutka

-- fixed bugs in FontData (reimplemented) and taking over (any objections michal?)
-- removed mutable as it is not needed

2008-11-10 09:34:37 Tree
[2e9a57] by Jozef Mišutka Jozef Mišutka

-- fix resource leak

2008-11-10 09:21:44 Tree
[eaedc4] by Michal Hocko Michal Hocko

Delinearizator inherits from CXRef and so it can't deallocate inputStream
and can't close file in the Delinearizator::getInstance fail (exception)
path because this is already done by ~CXRef called in the exception unwind
path from the constructor which throws an exception!

2008-11-07 13:46:37 Tree
[64a24d] by Jozef Mišutka Jozef Mišutka

-- synchro with newest changes

2008-11-07 12:35:12 Tree
[186d7a] by Jozef Mišutka Jozef Mišutka

-- zlib checkers

2008-11-07 12:33:39 Tree
[aae1a7] by Michal Hocko Michal Hocko

*** empty log message ***

2008-11-05 21:18:09 Tree
[be54f4] by Michal Hocko Michal Hocko

Subject: [PATCH @num@/@total@] version manipulation in gui updated

This patch changes implementation of version handling in gui. We don't rely
on hardcoded version in src/gui/version.h, but we use the output from
getversion script instead.

The second big change is that we don't have version defined as a macro in
the header file. Macro approach has one big disadvantage that we have to
rebuild all sources which use it each time when the version is changed.

In this patch we have changed VERSION to the extern variable defined in the module which is generated on demand by Makefile.gui. TARGET
binary depends on this cc module and Makefile.gui takes care that it
contains the current full version value (from getversion -v -r). File is
generated during first compilation and whenever some of the version
important files (getversion -f) is modified.

Implementation of cc module doesn't have above macro (rebuild all sources
including version.h) problem because the real value of extern variable for
version is used in the link time. So only necessary minimum is recompiled.

* dist-exclude, version_exclude
- is generated and shouldn't be distributed
- template for added
- this file is used by Makefile.gui to generate
- currently only defines PDFEDIT_VERSION extern variable
* Makefile.gui
- generates - simply substitute version with the current
FULL_VERSION (./getversion -v -r). Generation target depends on
all version important files
- TARGET depends on
* version.h
- VERSION -> PDFEDIT_VERSION (for disambiguate)
- PDFEDIT_VERSION is variable rather than macro
* gui/, gui/, gui/, gui/
- sync with above changes

2008-10-30 12:28:12 Tree
Older >

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

No, thanks