TestaSkin, a new tool on the road

[CrazyEddie]

martignasse wrote:
is the alias system in sync with the falagard system ?

i mean, it's like the alias system was looking for primary type (Falagard, WindowLook, TaharezLook) and not at the mapped one (who are introduced with the falagard system, if i understand correctly).

You have found an inconsistency bug Thanks for the find! I will fix this today.

CE.

[martignasse]

I will fix this today

cool, by fixing it, you mean in the CVS HEAD ? in this case, what about the 0.4 version ?

because i wanted to stay with 0.4 until the official 0.5 release

For more info, after some research in 0.4 source code, WindowFactoryManager.cpp :

Code: Select all

/*************************************************************************
   Returns true if a WindowFactory (or an alias) for a specified type
   is present
*************************************************************************/
bool WindowFactoryManager::isFactoryPresent(const String& name) const
{
   // first try for a 'real' type
   if (d_factoryRegistry.find(name) != d_factoryRegistry.end())
   {
      return true;
   }
   // no real type exists with that type, see if we have an appropriate alias instead.
   else
   {
      return (d_aliasRegistry.find(name) != d_aliasRegistry.end());
   }

}


this method isn't looking at falagard mapped type, and it's what causing trouble, appearently.
but maybe there is other miss i didn't noticed.

[CrazyEddie]

Hi,

This has now been fixed. It may take up to 24 hours for the anonymous cvs server to sync with the developer server.

Bug fixes generally go into the current stable branch, and then get merged back into head, and this is what I have done in this case. So for cases like these you just do a 'cvs update' on your 0.4 directory to get all fixes applied to that branch; this does not move you to head - you will stay on the stable branch.

The fix was slightly more involved than the one method (unfortunately ). The fixes applied should also solve a couple of other similar, potential issues. In doing the fix, creation semantics have changed slightly; aliases now take precedence over concrete types with the same name, which allows 'hiding' of old names via the aliasing system - which was always the intended behaviour (there are posts in here to that effect, though nobody every reported that it didn't work ).

HTH

CE.

[martignasse]

This has now been fixed. It may take up to 24 hours for the anonymous cvs server to sync with the developer server.

thanks thousand time

I'll look tomorrow.

[martignasse]

ok, i make it work but...found another bug

Code: Select all

// in CEGUIWindowFactoryManager.cpp, line 299
String WindowFactoryManager::getDereferencedAliasType(const String& type) const
{
    TypeAliasRegistry::const_iterator alias = d_aliasRegistry.find(type);

    // if this is an aliased type, ensure to fully dereference by recursively
    // calling ourselves on the active target for the given type.
    if (alias != d_aliasRegistry.end())
 |-|-|->   return getDereferencedAliasType(type);    <-|-|-|

    // we're not an alias, so return the input type unchanged
    return type;
}

this line make a dead lock when invoked with an alias type.

I modified like this to make it good for me

Code: Select all

String WindowFactoryManager::getDereferencedAliasType(const String& type) const
{
     TypeAliasRegistry::const_iterator alias = d_aliasRegistry.find(type);

    // if this is an aliased type, ensure to fully dereference by recursively
    // calling ourselves on the active target for the given type.
    if (alias != d_aliasRegistry.end())
      {
|-|-|->         return getDereferencedAliasType((*alias).second.getActiveTarget());  <-|-|-|
      }

    // we're not an alias, so return the input type unchanged
    return type;
}

but not sure about the validity of my correction.

notice the thread about the ceguistring.h(5550) compile error

About TestaSkin, just to say skin choice is ok to all layout type now
Just finish implementing it and i'll make an update

[CrazyEddie]

Hi,

Both of these issues were already fixed earlier this afternoon, and should be available in anonymous cvs any time soon

[martignasse]

ok, cool

[martignasse]

ok, finally got it

Always the same link for the update.

new in this release :
- A little toolbox to choice the background color.
- Layout Selector redesign, with a more confortable management of xml layout files
- You can now change the skin when you want, with both layout type

Feedback are greatly appreciated.

Have fun.

[martignasse]

and like always, a little screenshot :

[martignasse]

One important thing for those who want to test there own xml layout :

widgets type should be specified without skin. for example :
"Button" instead of "WindowsLook/Button".

It's to make it work with the Alias system.

for a complete list of widget type name, look in one of the WidgetAliasesXXXX.scheme file.

should begin to make a Doc, or a wiki page.

[martignasse]

come on, no feedback ?

nobody tried it ?

please, i need some feedback

[lindquist]

I'm on Linux, and due to lack of makefiles and patience to create some I tried running it through WINE 9.1 (win32 emu).

To my surprise it actually worked
Looks great, I just had to copy 'msvcr71.dll' and 'msvcp71.dll' from my windows partition to get it going.

I really like your marti skin. very clean

I got an 'unknown exception occured' dialog when I tried to add a layout from a file. (add button Clicked).
Have'nt looked into it, but besides that it worked great.

Good job

[spannerman]

I just tested your app, martignasse, works well, good job. I can see this being really useful, escpecially for newcomers deciding what skin to use in their GUI, but also for skin creators to quickily test and compare their own creations.

I have one suggestion: When switching skins, the layout (size, positioning etc) gets reset...is this a necessity? I think that it would be even better if the layout looked the same no matter what skins you switch to, and regardless of how the user has modified the layout.

Keep it up

[martignasse]

thanks guys...

to lindquist :

I'm on Linux, and due to lack of makefiles and patience to create some I tried running it through WINE 9.1 (win32 emu).

To my surprise it actually worked
Looks great, I just had to copy 'msvcr71.dll' and 'msvcp71.dll' from my windows partition to get it going.

good to know it's working with a win emulator

I really like your marti skin. very clean

yes, it's like y wanted it to be... but far from finished...working on

I got an 'unknown exception occured' dialog when I tried to add a layout from a file. (add button Clicked).

it's the one thing i wanted feedback on.
i have the same, completly randomised bug, and can't find where it come from


from spannerman :

I can see this being really useful, escpecially for newcomers deciding what skin to use in their GUI, but also for skin creators to quickily test and compare their own creations

yes, it's the idea behind this little tool (appart the fact i'm hardly learning CEGUI)

When switching skins, the layout (size, positioning etc) gets reset...is this a necessity? I think that it would be even better if the layout looked the same no matter what skins you switch to, and regardless of how the user has modified the layout.

i'm agree with you but the fact is, to switch between skin, i have to reload the layout. And the only way i see to keep things in place is to track every position and size....not a little job...not for twomorrow. but a good idea indeed[archived].

For now, next step is to make user skin add in.
and saving choice list in config file.

thanks again for feedback.

[Viper]

Martignasse,
I had taken some time away from working on my game using CEGUI, but I'm back now. This is an excellent program, and extremely useful. Keep up the excellent work.
I did notice that the selection of the skin did not seem to work correctly. When I selected, it used the one above the one that I clicked on.
I look forward to the finished product.