On Wednesday, 29 May 2013 at 09:51:04 UTC, Jacob Carlborg wrote:
On 2013-05-28 23:25, Flamaros wrote:
Hi,
I and a friend are developing a GUI library, and now our
script engine
is ready to start a prototype (but far to be finished). We
think to try
to create a GUI editor based on our library.
In this way, we'll see which features are need.
My concern is about how the editor have to works, we see two
different
ways to do it :
1) Classic editor external to the user applications
a) Good :
- Lightweight (easy to deploy and test)
- No need to modify application code
- Stable due to isolation of application
- Real-time edition but limited on one view (bad to
preview menus
transitions)
b) Bad :
- Limited, plugins needed to extend editor components and
his
knowledge of application (can't predict size of unknown
application
specific items)
2) Integrated editor (launch with the user application in a
second Window)
a) Good :
- Preview is the final result with real data
- All application components accessible to the editor
without
complex plugin system (in this way all editors components will
be well
placed in the preview)
- Full real-time edition (can preview menus
transitions,...)
- User can customize the editor
b) Bad :
- Intrusive in the application code
- Force the user to port application on a desktop OS
(Linux, Mac or
Windows), not friendly if he target only embedded devices (can
be
bypassed with a remote system)
- Less stable editor?
The second solution is commonly used in the video game
industry, but is
the best choice for a larger usage?
What do you think about?
I would go with the first approach because I would guess it's
easier. The editor creates the controls. When saving it will
serialize all the controls to some format. This format is then
read by the application.
For serialization you could have a look at Orange:
https://github.com/jacob-carlborg/orange
You think it's easier to do or to use?
We can't do serialization because our GUI files are lua scripts.
It looks like :
[CODE]
Image {
id = "image",
source = "images/pngtest.png",
x = 50,
y = 50,
titi = 0,
toto = function()
return image.width + image.height
end,
onTotoChanged = function()
image.titi = image.toto
print("onTotoChanged = "..image.titi)
end,
Button {
width = function()
return image.width
end,
height = function()
return image.height
end,
}
}
[/CODE]
Button isn't a D Object :
[CODE]
function Button(t)
print(t.width)
local Buttonimage3 = Image {
id = "Buttonimage",
width = 100, -- property binding, when image change width is
automatically updated
height = 50,
-- opGet = function(propertyName)
-- return ButtonmouseArea[propertyName]
-- end,
-- opSet = function(propertyName, value)
-- ButtonmouseArea[propertyName] = value
-- end,
source = function ()
if ButtonmouseArea.pressed then -- property binding, when
mouseArea state change this condition is updated directly
return "images/Alpha-blue-trans.png"
else
return "images/pngtest.png"
end
end,
MouseArea { -- parent/child
object encapsulation
id = "ButtonmouseArea",
width = function ()
return Buttonimage.width
end,
height = function ()
return Buttonimage.height
end,
},
}
for key, value in pairs(t) do
print(value)
Buttonimage3[key] = value
end
return Buttonimage3
end
[/CODE]
