kicad-developers team mailing list archive

[jp charras <jp.charras@xxxxxxxxxx>]

Le 12/10/2016 à 14:53, Tomasz Wlostowski a écrit :
> On 12.10.2016 14:49, jp charras wrote:
>> Hi Orson and Tomasz,
>>
>> A few days ago I had a look at this issue, previously reported to the Kicad forum.
>> Although it happens on 4.04 and not in the current version, I am thinking it is not fixed (it does
>> not happen just because there are a few changes which generate a slightly different zone shape).
>>
>> It happens because in SHAPE_POLY_SET::fractureSingle( POLYGON& paths ) this code:
>>
>>     // keep connecting holes to the main outline, until there's no holes left...
>>     while( num_unconnected > 0 )
>>     {
>> <skipped>
>>         num_unconnected -= processEdge( edges, smallestX );
>>     }
>> hangs if processEdge returns 0.
>>
>> processEdge can return 0, and this is the case with the minimal board I attached.
>>
>> May I ask you to have a look into this suspicious code (I am not familiar with it)?
>>
>> Thanks.
>>
> Hi JP,
> 
> Does the file outline created by 4.0.4 contain duplicate vertices?
> 
> Tom
> 

I don't know (not really easy to say).

Have a look at (in zones_convert_brd_items_to_polygons_with_Boost.cpp, kicad 4.04), line 456:
    fractured.Fracture();
It does not use the fast mode, and therefore the polygons are made strictly simple before fracturing
(unless I missed something).
and it hangs.

If modified like this
    fractured.Fracture( true );
It uses the fast mode, and therefore the polygons are not made strictly simple before fracturing,
but Fracture does not hang (perhaps just because the geometry has just changed a bit).

Note also the current code (in master) uses always the fast mode.


-- 
Jean-Pierre CHARRAS