wxruby users - Sep 2004 - wxruby-swig problem: MDIChildFrame::Maximize

I just grabbed the latest wxruby-swig from cvs and tried to compile it 
(without enabling the SWIG step) on my gentoo Linux box. I get:

g++  -c -I/usr/lib/wx/include/gtk2-2.4 -DGTK_NO_CHECK_CASTS -D__WXGTK__ 
-D_FILE_OFFSET_BITS=64 -D_LARGE_FILES  -O2 -march=athlon-xp  -fPIC   -I 
/usr/lib/ruby/gems/1.8/gems/rake-0.4.8/bin -I 
/usr/lib/ruby/gems/1.8/gems/rake-0.4.8/lib -I 
/usr/lib/ruby/site_ruby/1.8 -I /usr/lib/ruby/site_ruby/1.8/i686-linux -I 
/usr/lib/ruby/site_ruby -I /usr/lib/ruby/1.8 -I 
/usr/lib/ruby/1.8/i686-linux -I . -o obj/MDIChildFrame.o 
src/MDIChildFrame.cpp
src/MDIChildFrame.cpp: In function `VALUE 
_wrap_wxMDIChildFrame_Maximize(int,
    VALUE*, long unsigned int)'':
src/MDIChildFrame.cpp:921: error: no matching function for call to `
    wxMDIChildFrame::Maximize()''
/usr/include/wx/gtk/mdi.h:174: error: candidates are: virtual void
    wxMDIChildFrame::Maximize(bool)
rake aborted!

It seems that under wxgtk 2.4.2, the wx declaration is Maximize(bool), 
with a non-optional parameter. Under MSW or Mac is it an optional 
parameter, or no parameter at all?

Kevin

I''ve been bouncing between Mac and Windows, and hadn''t gotten
to trying
GTK quite yet. I am not surprised there are a few missing methods. I''ve
found that missing methods one one platform need to be removed on all 
platforms, because there is not a way to tell SWIG to wrap a method with 
the proper ifdefs. Post parsing would be really kludgy too, because you 
need to know a little too much info.

There are some classes (like wxToggleButton) that don''t exist on some 
platforms. If you add the following to the interface file

%runtime %{
//@@ifdef __WXMSW__
%}

fixplatform will replace the //@@ with a #, and add a second 
Init_wxWhatever method to the end of the sourcefile for all other 
platforms.

Nick


Kevin Smith wrote:
> I just grabbed the latest wxruby-swig from cvs and tried to compile it 
> (without enabling the SWIG step) on my gentoo Linux box. I get:
>
> g++  -c -I/usr/lib/wx/include/gtk2-2.4 -DGTK_NO_CHECK_CASTS 
> -D__WXGTK__ -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES  -O2 
> -march=athlon-xp  -fPIC   -I 
> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/bin -I 
> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/lib -I 
> /usr/lib/ruby/site_ruby/1.8 -I /usr/lib/ruby/site_ruby/1.8/i686-linux 
> -I /usr/lib/ruby/site_ruby -I /usr/lib/ruby/1.8 -I 
> /usr/lib/ruby/1.8/i686-linux -I . -o obj/MDIChildFrame.o 
> src/MDIChildFrame.cpp
> src/MDIChildFrame.cpp: In function `VALUE 
> _wrap_wxMDIChildFrame_Maximize(int,
>    VALUE*, long unsigned int)'':
> src/MDIChildFrame.cpp:921: error: no matching function for call to `
>    wxMDIChildFrame::Maximize()''
> /usr/include/wx/gtk/mdi.h:174: error: candidates are: virtual void
>    wxMDIChildFrame::Maximize(bool)
> rake aborted!
>
> It seems that under wxgtk 2.4.2, the wx declaration is Maximize(bool), 
> with a non-optional parameter. Under MSW or Mac is it an optional 
> parameter, or no parameter at all?
>
> Kevin
> _______________________________________________
> wxruby-users mailing list
> wxruby-users@rubyforge.org
> http://rubyforge.org/mailman/listinfo/wxruby-users
>
>
>

To answer your question, wxMac has Maximize() and Maximize(bool), but 
the second just calls the first. I suggest ignoring Maximize(), but 
forcing Maximize(bool) to have a default arg using extractxml.

Nick

Kevin Smith wrote:
> I just grabbed the latest wxruby-swig from cvs and tried to compile it 
> (without enabling the SWIG step) on my gentoo Linux box. I get:
>
> g++  -c -I/usr/lib/wx/include/gtk2-2.4 -DGTK_NO_CHECK_CASTS 
> -D__WXGTK__ -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES  -O2 
> -march=athlon-xp  -fPIC   -I 
> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/bin -I 
> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/lib -I 
> /usr/lib/ruby/site_ruby/1.8 -I /usr/lib/ruby/site_ruby/1.8/i686-linux 
> -I /usr/lib/ruby/site_ruby -I /usr/lib/ruby/1.8 -I 
> /usr/lib/ruby/1.8/i686-linux -I . -o obj/MDIChildFrame.o 
> src/MDIChildFrame.cpp
> src/MDIChildFrame.cpp: In function `VALUE 
> _wrap_wxMDIChildFrame_Maximize(int,
>    VALUE*, long unsigned int)'':
> src/MDIChildFrame.cpp:921: error: no matching function for call to `
>    wxMDIChildFrame::Maximize()''
> /usr/include/wx/gtk/mdi.h:174: error: candidates are: virtual void
>    wxMDIChildFrame::Maximize(bool)
> rake aborted!
>
> It seems that under wxgtk 2.4.2, the wx declaration is Maximize(bool), 
> with a non-optional parameter. Under MSW or Mac is it an optional 
> parameter, or no parameter at all?
>
> Kevin
> _______________________________________________
> wxruby-users mailing list
> wxruby-users@rubyforge.org
> http://rubyforge.org/mailman/listinfo/wxruby-users
>
>
>

OK. I just checked in new wxruby-swig code.

It now uses REXML instead of NQXML. I updated the README accordingly.

To get it to compile under Linux, I had to add platform-specific 
%ignores for a handful of methods. I was able to do this in the .i 
files, but I would like to find a slightly cleaner way. I have an idea 
that I need to try.

I also had to completely remove a couple classes (Notebook and Pen) from 
the linux build, using Nick''s //@@ifdef trick. They each had a virtual 
method that doesn''t exist under linux. Instead of removing the classes 
completely, I think I can %extend the class to include the missing 
method (under linux only), and then %ignore the method (again, linux-only).

Also, I just saw the Event stuff you (Nick) added. Can you briefly 
describe what it is doing?

Kevin


Nick wrote:
> 
> I''ve been bouncing between Mac and Windows, and hadn''t
gotten to trying
> GTK quite yet. I am not surprised there are a few missing methods.
I''ve
> found that missing methods one one platform need to be removed on all 
> platforms, because there is not a way to tell SWIG to wrap a method with 
> the proper ifdefs. Post parsing would be really kludgy too, because you 
> need to know a little too much info.
> 
> There are some classes (like wxToggleButton) that don''t exist on
some
> platforms. If you add the following to the interface file
> 
> %runtime %{
> //@@ifdef __WXMSW__
> %}
> 
> fixplatform will replace the //@@ with a #, and add a second 
> Init_wxWhatever method to the end of the sourcefile for all other 
> platforms.
> 
> Nick
> 
> 
> Kevin Smith wrote:
> 
>> I just grabbed the latest wxruby-swig from cvs and tried to compile it 
>> (without enabling the SWIG step) on my gentoo Linux box. I get:
>>
>> g++  -c -I/usr/lib/wx/include/gtk2-2.4 -DGTK_NO_CHECK_CASTS 
>> -D__WXGTK__ -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES  -O2 
>> -march=athlon-xp  -fPIC   -I 
>> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/bin -I 
>> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/lib -I 
>> /usr/lib/ruby/site_ruby/1.8 -I /usr/lib/ruby/site_ruby/1.8/i686-linux 
>> -I /usr/lib/ruby/site_ruby -I /usr/lib/ruby/1.8 -I 
>> /usr/lib/ruby/1.8/i686-linux -I . -o obj/MDIChildFrame.o 
>> src/MDIChildFrame.cpp
>> src/MDIChildFrame.cpp: In function `VALUE 
>> _wrap_wxMDIChildFrame_Maximize(int,
>>    VALUE*, long unsigned int)'':
>> src/MDIChildFrame.cpp:921: error: no matching function for call to `
>>    wxMDIChildFrame::Maximize()''
>> /usr/include/wx/gtk/mdi.h:174: error: candidates are: virtual void
>>    wxMDIChildFrame::Maximize(bool)
>> rake aborted!
>>
>> It seems that under wxgtk 2.4.2, the wx declaration is Maximize(bool), 
>> with a non-optional parameter. Under MSW or Mac is it an optional 
>> parameter, or no parameter at all?
>>
>> Kevin
>> _______________________________________________
>> wxruby-users mailing list
>> wxruby-users@rubyforge.org
>> http://rubyforge.org/mailman/listinfo/wxruby-users
>>
>>
>>
> 
> _______________________________________________
> wxruby-users mailing list
> wxruby-users@rubyforge.org
> http://rubyforge.org/mailman/listinfo/wxruby-users

> OK. I just checked in new wxruby-swig code.
>
> It now uses REXML instead of NQXML. I updated the README accordingly.
>
Cool cool. Does this speed up extractxml at all?
> To get it to compile under Linux, I had to add platform-specific 
> %ignores for a handful of methods. I was able to do this in the .i 
> files, but I would like to find a slightly cleaner way. I have an idea 
> that I need to try.
>
> I also had to completely remove a couple classes (Notebook and Pen) 
> from the linux build, using Nick''s //@@ifdef trick. They each had
a
> virtual method that doesn''t exist under linux. Instead of removing
the
> classes completely, I think I can %extend the class to include the 
> missing method (under linux only), and then %ignore the method (again, 
> linux-only).
The problem is finding a way to get the generated source code to compile 
on *all* platforms, instead of needing to generate Linux, Mac. and 
Windows "versions" of the source. I had some Mac specific
%ignore''s in
there (and created a $swig_options variable that was overridable by the 
rake<platform>, but I realized that that generates platform specific 
source, which goes against the goal of source that compiles on all 
platforms. I am very curious how wxPython gets around this.
>
> Also, I just saw the Event stuff you (Nick) added. Can you briefly 
> describe what it is doing?
Certainly. Basically, I got sick of hand-writing all those event 
handlers, so upon some investigation I found the event definitions 
already existed in the XML file. I use the argument signature to map to 
the appropriate call type (using straight string matching, which I''ll 
admit is an incredibly stupid way), and I scan the textual information 
to see if it defines the wxEVT_XXX that needs to be filled in. Of 
course, this isn''t in a lot of the decriptions, so I have a moderate 
sized number of corrections in extractxml. All of this goes into 
''swig/classes/include/events.rb''

I moved a lot of the basic functions of EvtHandler.i into Events.i. The 
script fixevents.rb reads events.rb and uses the event declarations to 
generate the C++ code for the event handlers. It currently wraps Windows 
only events in the proper ifdefs. This only handles the event handlers, 
so the actual event mappings still need to be done in EvtHandler.i. This 
does take a huge weight off somebody adding a new class, because we 
already support their event handlers. They just need to add the event 
class definition and the mapping.

All of this was added to the tutorial page.

Nick


Kevin Smith wrote:
> OK. I just checked in new wxruby-swig code.
>
> It now uses REXML instead of NQXML. I updated the README accordingly.
>
> To get it to compile under Linux, I had to add platform-specific 
> %ignores for a handful of methods. I was able to do this in the .i 
> files, but I would like to find a slightly cleaner way. I have an idea 
> that I need to try.
>
> I also had to completely remove a couple classes (Notebook and Pen) 
> from the linux build, using Nick''s //@@ifdef trick. They each had
a
> virtual method that doesn''t exist under linux. Instead of removing
the
> classes completely, I think I can %extend the class to include the 
> missing method (under linux only), and then %ignore the method (again, 
> linux-only).
>
> Also, I just saw the Event stuff you (Nick) added. Can you briefly 
> describe what it is doing?
>
> Kevin
>
>
> Nick wrote:
>
>>
>> I''ve been bouncing between Mac and Windows, and
hadn''t gotten to
>> trying GTK quite yet. I am not surprised there are a few missing 
>> methods. I''ve found that missing methods one one platform need
to be
>> removed on all platforms, because there is not a way to tell SWIG to 
>> wrap a method with the proper ifdefs. Post parsing would be really 
>> kludgy too, because you need to know a little too much info.
>>
>> There are some classes (like wxToggleButton) that don''t exist
on some
>> platforms. If you add the following to the interface file
>>
>> %runtime %{
>> //@@ifdef __WXMSW__
>> %}
>>
>> fixplatform will replace the //@@ with a #, and add a second 
>> Init_wxWhatever method to the end of the sourcefile for all other 
>> platforms.
>>
>> Nick
>>
>>
>> Kevin Smith wrote:
>>
>>> I just grabbed the latest wxruby-swig from cvs and tried to compile
>>> it (without enabling the SWIG step) on my gentoo Linux box. I get:
>>>
>>> g++  -c -I/usr/lib/wx/include/gtk2-2.4 -DGTK_NO_CHECK_CASTS 
>>> -D__WXGTK__ -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES  -O2 
>>> -march=athlon-xp  -fPIC   -I 
>>> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/bin -I 
>>> /usr/lib/ruby/gems/1.8/gems/rake-0.4.8/lib -I 
>>> /usr/lib/ruby/site_ruby/1.8 -I 
>>> /usr/lib/ruby/site_ruby/1.8/i686-linux -I /usr/lib/ruby/site_ruby
-I
>>> /usr/lib/ruby/1.8 -I /usr/lib/ruby/1.8/i686-linux -I . -o 
>>> obj/MDIChildFrame.o src/MDIChildFrame.cpp
>>> src/MDIChildFrame.cpp: In function `VALUE 
>>> _wrap_wxMDIChildFrame_Maximize(int,
>>>    VALUE*, long unsigned int)'':
>>> src/MDIChildFrame.cpp:921: error: no matching function for call to
`
>>>    wxMDIChildFrame::Maximize()''
>>> /usr/include/wx/gtk/mdi.h:174: error: candidates are: virtual void
>>>    wxMDIChildFrame::Maximize(bool)
>>> rake aborted!
>>>
>>> It seems that under wxgtk 2.4.2, the wx declaration is 
>>> Maximize(bool), with a non-optional parameter. Under MSW or Mac is 
>>> it an optional parameter, or no parameter at all?
>>>
>>> Kevin
>>> _______________________________________________
>>> wxruby-users mailing list
>>> wxruby-users@rubyforge.org
>>> http://rubyforge.org/mailman/listinfo/wxruby-users
>>>
>>>
>>>
>>
>> _______________________________________________
>> wxruby-users mailing list
>> wxruby-users@rubyforge.org
>> http://rubyforge.org/mailman/listinfo/wxruby-users
>
>
> _______________________________________________
> wxruby-users mailing list
> wxruby-users@rubyforge.org
> http://rubyforge.org/mailman/listinfo/wxruby-users
>
>
>