MAN_SUBDIR parameter

sqlite3 fixes

Add line numbers in case comment is not closed properly.

Enabling possibility to have { and } in (latex) index items

This verifies the 'memberdef' table against the generated XMLs.
It accomplishes this by executing a sql query, constructed by iterating
through the children and attributes of the memberdef XML element.

The information translated from xml to sql, is then removed from the DOM,
and the unprocessed bits are printed as warning messages.

usage:
    testing/testsqlite3.py -d /.../doxygen_sqlite3.db -x /.../xml/test__x11_8cpp.xml

output:
WARNING: 'location' has unprocessed attr 'bodyend'
WARNING: 'memberdef' has unprocessed child elem 'briefdescription'
WARNING: 'memberdef' has unprocessed child elem 'detaileddescription'
WARNING: 'memberdef' has unprocessed child elem 'inbodydescription'
WARNING: 'memberdef' has unprocessed child elem 'location'
WARNING: 'memberdef' has unprocessed attr 'kind'
WARNING: 'memberdef' has unprocessed attr 'prot'
WARNING: 'memberdef' has unprocessed attr 'static'
WARNING: 'memberdef' has unprocessed attr 'id'

In table basecompoundref, use SQLITE_TRANSIENT for
base, derived, refid

memberdef table changes:
* rename bstart to bodystart
* rename id_bfile to id_bodyfile
* add bodyend
* add {brief,detailed,inbody}description
Signed-off-by: Adrian Negreanu <adrian.m.negreanu@intel.com>

sqlite3: fedora has libsqlite3.so in /usr/lib64

Signed-off-by: Adrian Negreanu <groleo@gmail.com>

This bug fix is an extension to the bug fix:  Add warning when encountering a nested comment start (/*) without matching end (*/).
It is quite hard to find the place in the source file where the problem occurs. This bug fix helps to locate part of the problems, it is not perfect it only gives the starting comment tags that give a problem (other problems also occur e.g. when // is followed by */).

Bug 722759 - Documentation for derived classes no longer has an "Additional Inherited Members" section after upgrading Doxygen from 1.8.3.1 to 1.8.6

Incorrect handling dependencies

The handling of a number of dependencies was incorrect, it was stated that the i(source) cpp file depended on the (included) h file, this should be that the resulting (object) o file depends on the cpp and the h file. See e.g. the layout.cpp dependency
Some dependencies used ../generated_src/doxygen it is better to use the definition GENERATED_SRC

This patch is a patch for the build system so that less (no) generated files are in the source directories and also that the different build (on windows are independent of each others). Also the different sub-projects (e.g. doxygen , doxywizard) don't share any files anymore.

- ./.gitignore
  A number of files are not in the src directory anymore. generated directories are included. rtf and xml directories are there for possible future use.

- ./Doxyfile
  The generated files are not anymore in the src directory so they don't have to be excluded anymore

- ./configure
  Adjustmenst so the generated files are in the different subdirectories

- ./Makefile.in
- addon/doxmlparser/examples/metrics/metrics.pro.in
- addon/doxmlparser/src/doxmlparser.pro.in
- addon/doxmlparser/test/xmlparse.pro.in
- addon/doxyapp/doxyapp.pro.in
- addon/doxysearch/doxyindexer.pro.in
- addon/doxysearch/doxysearch.pro.in
- addon/doxywizard/Makefile.in
- addon/doxywizard/doxywizard.pro.in
- libmd5/libmd5.pro.in
- qtools/qtools.pro.in
- src/Makefile.in
- src/doxygen.pro.in
- src/libdoxycfg.pro.in
- src/libdoxycfg.t.in
- src/libdoxygen.pro.in
- src/libdoxygen.t.in
  Adjustment to the new directory structure

- addon/doxywizard/expert.cpp
  include file is generated in the generated_src/doxywizard directory, referenced through -I on compile line

- src/lang_cfg.h
  File is now automatically generated

- src/languages.py
  Correction of used path for new structure

- src/settings.py
- src/version.py
  Place for resulting file handled through an argument, so it is possible to have a separate file for each sub-project

- tmake/lib/win32-g++/generic.t
- tmake/lib/win32-mingw/generic.t
  The destination path was added twice, in case of a depth of more than 1 level (doxmlparser) this lead to problems

- winbuild/Config.rules
- winbuild/Gen_head.rules
- winbuild/Languages.rules
- winbuild/Lex.rules
- winbuild/Settings.rules
- winbuild/Version.rules
  correcting some path of generated files

- winbuild/Doxygen.sln
  Build order dependency could be dropped  as the version.cpp file is now a generated file for doxygen and doxywizard

- winbuild/Doxygen.vcproj
- winbuild/Doxywizard.vcproj
  Corrected paths to the new structure. Added unistd as an external generated file

- winbuild/unistd.h
  File is now generated

- winbuild/Unistd.rules
- winbuild/unistd.py
  Generated unistd for doxygen and doxywizard in a consistent way.

- winbuild/doxyindexer.vcproj
- winbuild/doxysearch.vcproj
- winbuild/qtools.vcproj
  Corrected paths to the new structure.

- winbuild/runbison.bat
- winbuild/version.bat
  Not used anymore everything handled through rules.

Documentation corrections

Fix to VHDL scanner.

The following code should produce two groups, each containing one page, within the same documentation scope,
currently it nests these groups and pages, resulting in incorrect documentation.

--! \defgroup group1 Group 1
--! \{
--!   @brief Group 1
--! \}

--! \ingroup group1
--! \{
--! \page group1_page1 Page 1
--! \verbatim
--!  Some Page 1 text
--! \endverbatim
--! \}

--! \defgroup group2 Group 2
--! \{
--!   @brief Group 2
--! \}

--! \ingroup group2
--! \{
--! \page group2_page1 Page 1
--! \verbatim
--!  Some Page 1 text
--! \endverbatim
--! \}

Similar C code to demonstrate the correct output.

//! \defgroup group1 Group 1
//! \{
//!   @brief Group 1
//! \}

//! \ingroup group1
//! \{
//! \page group1_page1 Page 1
//! \verbatim
//!  Some Page 1 text
//! \endverbatim
//! \}

//! \defgroup group2 Group 2
//! \{
//!   @brief Group 2
//! \}

//! \ingroup group2
//! \{
//! \page group2_page1 Page 1
//! \verbatim
//!  Some Page 1 text
//! \endverbatim
//! \}

Wrong UTF 8 codes

In the doxygen manual the index items for \{ and \} were missing due to the missing support for the usage of { and } in parts of the code.
This patch fixes this problem by introducing 2 new latex commands. See also http://tex.stackexchange.com/questions/153291/index-unmatched-braces-in-latex
Further improvements in the index are:
- consistency in different places
- correction of index for \::
- placing \~ on a more logical place (together with other special characters, ~ is in the ASCII table after a-z whilst other characters are before this range)

The utf-8 code for eta was identical to the code for theta and the code for Prime was identical to the one for prime

sqlite3: clear bindings on errors and more care with return

* call sqlite3_clear_bindings when step fails
* more miss-placed returns
Signed-off-by: Adrian Negreanu <adrian.m.negreanu@intel.com>