kicad-developers team mailing list archive

[jp charras <jp.charras@xxxxxxxxxx>]

Le 17/06/2015 19:56, jp charras a écrit :
> Le 17/06/2015 19:35, Chris Pavlina a écrit :
>> You probably saw my thread about the bounding boxes for LIB_TEXT and 
>> LIB_FIELD, which were being merged with the main component bounding box 
>> incorrectly. I tracked down what the bug was:
>>
>> in ::GetBoundingBox(), the coordinates are inverted because "Y 
>> coordinates for LIB_ITEMs are bottom to top". Problem is, the bounding 
>> boxes aren't used that way, they're expected in most places to be 
>> noninverted. The only place the LIB_TEXT bounding box should be inverted 
>> is in ::drawGraphics.
>>
>> This patch moves the inversion to ::drawGraphics, fixing the bounding 
>> box bug. The overall bounding box is now calculated correctly:
>>
>> http://misc.c4757p.com/correct_bbox.png
>>
>> Compare to the original version, where the text object that I moved 
>> upwards caused the box to extend downwards:
>>
>> http://misc.c4757p.com/bbox_wtf.png
>>
>> This patch also corrects one other minor bug: the bounding box was 
>> always calculated to include (0,0), even if the component itself doesn't 
>> include this point.
>>
>> --
>> Chris
> 
> 
> Thanks, Chris.
> 
> I'll have a look at these issues, because during investigations, I found
> some other strange things.
> 
> You are right about incorrect Bounding box calculations:
> Some use a bottom to top Y axis, some other a top to bottom Y axis.
> This is the explanation of this issue.
> 
> 

Incorrect Bounding box calculations in Library editor should be fixed in
rev 5770.


-- 
Jean-Pierre CHARRAS