Re: [unclassified] patch for color_properties assignment

[Kai Habel]

kahacjde schrieb:
> kahacjde wrote:
>   
> > John W. Eaton schrieb:
> >     
> > > I'm now confused about the patches in this thread and the 
> > > "problem with radio values in color_property" thread.  Which ones are
> > > needed?  Am I missing any other patches for graphics?
> > >
> > > Thanks,
> > >
> > > jwe
> > >
> > >   
> > >       
> > I think there is still one problem unsolved. Here is the relevant part 
> > of the discussion
> > (http://www.nabble.com/patch-for-color_properties-assignment-tf4041176.html)
> >
> >     
> > > > Another problem is the radio_values::validate method. In your patch you
> > > > check first for valid radio values:
> > > >
> > > >           if (radio_val.validate (s))
> > > > +           {
> > > > +             current_val = s;
> > > > +             current_type = radio_t;
> > > > +           }
> > > > +           else
> > > > +           {
> > > > +             color_values col (s);
> > > > +             if (! error_state)
> > > > +               {
> > > > +                 color_val = col;
> > > > +                 current_type = color_t;
> > > > +               }
> > > > +             else
> > > > +               error ("invalid color specification");
> > > > +           }
> > > >
> > > >
> > > >
> > > > but if 's' is a string color value (e.g. 'r' or 'red')  validate calls
> > > > the error function and sets the error flag. This causes an error message
> > > > and octave terminates. I have turned off the error call in
> > > > radio_values::validate for test purposes. Then the creation of a line
> > > > object with:
> > > > h=__go_line__ (gca);
> > > >
> > > > works as expected.
> > > >         
> > > I think the issue here is that validate raises an error. I will send a
> > > patch which will overcome this (e,g, replace radio_val.validate (s)
> > > with
> > >
> > > radio_val.possible_vals.find (val) != radio_val.possible_vals.end ()
> > >       
> > I have tried to replace "radio_val.validate (s)" with 
> > "radio_val.possible_vals.find (val) != radio_val.possible_vals.end ()" 
> > for the operator
> >
> > color_property&
> > color_property::operator = (const octave_value& val)
> >
> > (line 187 of graphics.cc), but this causes an compilation error:
> >
> > g++ -c  -fPIC -I. -I.. -I../liboctave -I../src -I../libcruft/misc   
> > -DHAVE_CONFIG_H -mieee-fp -Wall -W -Wshadow -Wold-style-cast -g -O2 
> > graphics.cc -o pic/graphics.o
> > graphics.h: In member function 'color_property& 
> > color_property::operator=(const octave_value&)':
> > graphics.h:112: error: 'std::set<std::basic_string<char, 
> > std::char_traits<char>, std::allocator<char> >, 
> > std::less<std::basic_string<char, std::char_traits<char>, 
> > std::allocator<char> > >, std::allocator<std::basic_string<char, 
> > std::char_traits<char>, std::allocator<char> > > > 
> > radio_values::possible_vals' is private
> > (Please note I use a modified graphics.h, the line 112 refers to 
> > std::set<std::string> possible_vals;)
> >
> > I don't know if some 'friend' magic  would help here. Another solution 
> > is to have an addtional radio_values::__validate__() function without 
> > this error() call.
> >
> > I don't know how to proceed here, any suggestions?
> >
> > Kai
> >
> >
> >     
>
>
> O.k. I propose the following patches to make the color assigment work again.
> As you might have noticed currently something like:
> plot([0 1],'Color','blue') 
> is not possible because "color_property::operator = (const octave_value&
> val)" is to restrictive. 
>
> If you look at the code of this operator (around line 179 of graphics.cc) if
> 's' is an string the assigment must fail because either "validate(s)" or
> later "color_values col (s);" raise an error. 
>
> I have introduced a __validate__ function which just checks for valid
> entries.
>
> Here is the respective Changelog:
>
> 2007-09-01  Kai Habel  <address@hidden>
>
>         * graphics.h (radio_values::__validate__): New function.
>         * graphics.cc (color_property::operator =): Use __validate__ instead of
> validate.
>
> Kai
>
> P.S. I send this email via nabble so I hope the attachments are correct.
>
> http://www.nabble.com/file/p12443384/graphics.h.patch graphics.h.patch 
> http://www.nabble.com/file/p12443384/graphics.cc.patch graphics.cc.patch 
>   
O.k. since the patches are not attached, I attach them here

Kai

--- graphics.cc.orig    2007-09-01 20:52:06.000000000 +0000
+++ graphics.cc 2007-09-01 20:52:58.000000000 +0000
@@ -184,7 +184,7 @@
 
       if (! s.empty ())
        {
-         if (radio_val.validate (s))
+         if (radio_val.__validate__ (s))
            {
              current_val = s;
              current_type = radio_t;

--- graphics.h.orig     2007-09-01 20:52:09.000000000 +0000
+++ graphics.h  2007-09-01 20:54:18.000000000 +0000
@@ -79,7 +79,7 @@
   {
     bool retval = true;
 
-    if (possible_vals.find (val) == possible_vals.end ())
+    if (!__validate__ (val))
       {
        error ("invalid value = %s", val.c_str ());
        retval = false;
@@ -87,6 +87,12 @@
 
     return retval;
   }
+  
+  bool __validate__ (const std::string& val)
+  {
+    return (possible_vals.find (val) != possible_vals.end ());
+  }
+
 
 private:
   // Might also want to cache