display.newLine(sceneGroup, centerX, centerY, centerX, centerY) bug?

Solar2D General Questions/Discussion
#1

Dear Corona community,

I am implementing a drawing function and am struggling with display.newLine:

I am experiencing different behaviours on different Android devices when I initialize a newline like this:

lokal stroke = display.newLine(sceneGroup, centerX, centerY, centerX, centerY)

As soon as I append to that line by lets say

stroke.append(centerX + 10, centerY + 20)

I get two results:

Huawei P20 lite draws the expected line from center to that specified point PLUS a line from somewhere topleft out of screen to the center point.

Samsung Galaxy Tab 2 also draws the expected line fromc enter to that specified point PLUS a line from somewhere topleft out of screen to the center point.

Corona Simulator on windows just draws the expected line

Samsung S5 mini also just draws the expected line

So, I have no idea where this additional line comes from. I can solve this issue by setting up the initial newline like this:

lokal stroke = display.newLine(sceneGroup, centerX, centerY, centerX + 1, centerY + 1)

so it seems to me that setting up a newline with same start and target coordinates causes trouble on some devices.

Is this a bug or am I getting something wrong?

Thanks :slight_smile:

#2

Hi CoronaChris,

It seems you use it wrong. The append function requires colon notation e.g.

stroke:append( centerX + 10, centerY + 20 )

I’m not sure using the same point as start and end point is good idea.

Read more :

Have a nice day:)

ldurniat

#3

There is a bug in display.newLine…

#4

Note: You shouldn’t be creating a zero-length segement to start your line.

That is an incorrect practice.

#5

There may be a bug, but creating a 0 length line seems like it could be the source of the problem.

Rob

#6

Thank you all for your input.

@Idurniat: I just made a spelling mistake, in my real code its written with a colon :slight_smile:

@greg, roaminggamer and Rob: All right then, I’ll find a different way to initialize my newline. Guess in the end it’s some sort of bug, but I agree, technically a 0 length does not make that much sense…

#7

@CoronaChris -  I don’t agree.  It is not a bug.  It isn’t meant to be used that way so you’re getting an undefined behavior.  The bug is in your code and usage.

Note: My ‘strong’ response is because I am not fond of throwing around the word bug lightly.

#8

Sorry for stepping in here, I was referring to the Line Append problem, - it only happens when appending horizontal line segments.  All other non horizontal line segments add fine…

#9

@gregbrady - FYI (in case of confusion), I was not responding to you above. I was responding to CoronaChris saying … oh well it must be a Corona bug.

The premise was wrong though because he is using the feature incorrectly to begin with and any unexpected results are then because of this incorrect usage.

#10

@roaminggamer, i don’t agree with you. if the function let you define the value to 0 and behave wrong it’s a bug with that function, not the people that use it wrong. as a programmer when i do a function i always try to think about all the possibilities a user can do with that function, including bad practices, ofc it’s almost impossible to think of all, depending of the complexity of that function, so bugs appear time to time.

#11

Of course our API’s need to be hardened to protect against invalid input, so if you find a way that causes it to misbehave we should look at addressing it. But if it’s fixable on the user end by not doing the invalid input, file a bug report and work around the problem.

That said, we need a sample project that demonstrates the error. This should include at a minimum a main.lua, config.lua and build.settings and any assets needed to duplicate the problem. Put this in a .zip file. It needs a project that we can simply build and install on a device and see the error happen. Once you have that, use the “Report a bug” link at the top of the forum page to file the report.

Thanks

Rob

#12

I hear you and agree in part w/ what you’re saying, but at the same time I dont’ want to see any Corona functions over engineered to protect against things that don’t make sense and may cost a lot of cycles to detect and handle. i.e. I worry about the trade-off between robustness and performance.

Still. You make a good point. Sometimes zero (or near-zero) length segments can be introduced as part of a calculation on the user side and these are hard to catch, so having the SDK handle it would be nice.

#13

@roaminggamer, I understand and agree with your point of over engineering and making more cycles than we need to make a function work properly. we have to make those decisions every day. it will depend in each function and the sweet spot  is hard to get sometimes.

But, usually those measures to make the function more rebust, are neglectable. It’s far more important to focus and optimize the rest of the code than those measures that can break your code in a second. I revise my code all the time to get it smaller/faster every day I look at it.

#14

Hi CoronaChris,

It seems you use it wrong. The append function requires colon notation e.g.

stroke:append( centerX + 10, centerY + 20 )

I’m not sure using the same point as start and end point is good idea.

Read more :

Have a nice day:)

ldurniat

#15

There is a bug in display.newLine…

#16

Note: You shouldn’t be creating a zero-length segement to start your line.

That is an incorrect practice.

#17

There may be a bug, but creating a 0 length line seems like it could be the source of the problem.

Rob

#18

Thank you all for your input.

@Idurniat: I just made a spelling mistake, in my real code its written with a colon :slight_smile:

@greg, roaminggamer and Rob: All right then, I’ll find a different way to initialize my newline. Guess in the end it’s some sort of bug, but I agree, technically a 0 length does not make that much sense…

#19

@CoronaChris -  I don’t agree.  It is not a bug.  It isn’t meant to be used that way so you’re getting an undefined behavior.  The bug is in your code and usage.

Note: My ‘strong’ response is because I am not fond of throwing around the word bug lightly.

#20

Sorry for stepping in here, I was referring to the Line Append problem, - it only happens when appending horizontal line segments.  All other non horizontal line segments add fine…