Trying to extend the "display" library functions, works in simulator but fails on actual devices

Solar2D General Questions/Discussion
#1

Hello,

Recently I’ve had the need of extending the “display” library of Corona.

The thing I was trying to do is saving the r/g/b/a values of a displayObject as properties on the displayObject itself. See code below:

local defaultR, defaultG, defaultB, defaultA = display.getDefault("fillColor") local defaultLineR, defaultLineG, defaultLineB, defaultLineA = display.getDefault("lineColor") local function extendDisplayObject(\_funcName) local cachedFunc = display[\_funcName] display[\_funcName] = function(...) local obj = cachedFunc(unpack(arg)) obj.r = \_funcName == "newLine" and defaultLineR or defaultR obj.g = \_funcName == "newLine" and defaultLineG or defaultG obj.b = \_funcName == "newLine" and defaultLineB or defaultB obj.a = \_funcName == "newLine" and defaultLineA or defaultA local cachedSetFillColor = obj.setFillColor obj.setFillColor = function(...) obj.r = arg[2] obj.g = arg[3] obj.b = arg[4] obj.a = arg[5] or obj.a cachedSetFillColor(unpack(arg)) end return obj end end extendDisplayObject("newCircle") extendDisplayObject("newContainer") extendDisplayObject("newEmbossedText") extendDisplayObject("newEmitter") extendDisplayObject("newImage") extendDisplayObject("newImageRect") extendDisplayObject("newLine") extendDisplayObject("newPolygon") extendDisplayObject("newRect") extendDisplayObject("newRoundedRect") extendDisplayObject("newSnapshot") extendDisplayObject("newSprite") extendDisplayObject("newText")  

It works in the simulator, but on devices (iOS & Android) it gives the following error:

“…display.lua:16: attempt to index local ‘obj’ (a nil value)”

Could someone point me in the right direction to look for the solution?

#2

What version of Corona are you using?

Ex: The latest version/build is: 

2016.2879

#3

Works for me on version 2879.

#4

Which function is failing?

The only thing that stands out is that unpack could give you inconsistent results if arg contains nil s. But I’m hesitant, since I can’t think of any of those APIs not already balking at a nil argument. (Maybe the group?) Any better luck if you call it as

unpack(arg, 1, arg.n)

or do you get the same result?

Also, your setFillColor is broken for n = 1 or 2, assuming you intend Corona’s behavior. Off the top of my head, you probably want something like

if arg.n \<= 2 then if type(arg[1]) == "table" then -- Gradient... else obj.r, obj.g, obj.b = arg[1], arg[1], arg[1] obj.a = arg[2] or obj.a end else obj.r = arg[1] -- start at 1 obj.g = arg[2] obj.b = arg[3] obj.a = arg[4] or obj.a end
#5

Sorry for wasting your time, but the problem was not in the extension itself, but in the code that was calling this.

As we all know devices are case-sensitive and desktops (atleast in my case) are not…

Corona has implemented a very useful functionality that warns us for case-sensitive mistakes, but as I was logging to much myself, so I overlooked.

#6

What version of Corona are you using?

Ex: The latest version/build is: 

2016.2879

#7

Works for me on version 2879.

#8

Which function is failing?

The only thing that stands out is that unpack could give you inconsistent results if arg contains nil s. But I’m hesitant, since I can’t think of any of those APIs not already balking at a nil argument. (Maybe the group?) Any better luck if you call it as

unpack(arg, 1, arg.n)

or do you get the same result?

Also, your setFillColor is broken for n = 1 or 2, assuming you intend Corona’s behavior. Off the top of my head, you probably want something like

if arg.n \<= 2 then if type(arg[1]) == "table" then -- Gradient... else obj.r, obj.g, obj.b = arg[1], arg[1], arg[1] obj.a = arg[2] or obj.a end else obj.r = arg[1] -- start at 1 obj.g = arg[2] obj.b = arg[3] obj.a = arg[4] or obj.a end
#9

Sorry for wasting your time, but the problem was not in the extension itself, but in the code that was calling this.

As we all know devices are case-sensitive and desktops (atleast in my case) are not…

Corona has implemented a very useful functionality that warns us for case-sensitive mistakes, but as I was logging to much myself, so I overlooked.