← Back to team overview

kicad-developers team mailing list archive

Re: GAL canvas strategy - testers needed!


Le 19/09/2018 à 13:01, Rene Pöschl a écrit :
> On 19/09/2018 11:54, jp charras wrote:
>> Le 19/09/2018 à 11:40, Rene Pöschl a écrit :
>>> Ok i looked at the pot symbols. They are indeed strange. (ignore my last
>>> message)
>>> However, how does the new fill algorithm work with things like the logic
>>> gates. (Example the symbol for 74LS00) It has an open polygon on purpose
>>> as it consists of that polygon plus an arc. We still want the body do be
>>> filled but do not want a visible line between the arc and the polygon
>>> body.
>> There is no problem with 74LS00 and 74LS02 shapes.
>> You could find an issue with a 74LS32, but the polygon used as a filled
>> shape is broken: its line width is set to -1000 mils, a bit strange.
> The reason for that is that kicad did not (or does not) support filling
> shapes with concave arcs in it. So the outline was made with arcs and
> lines whereas the filled part was made with a single polygon that
> approximates the shape needed. To reduce the number of needed line
> segments the shape is approximated quite crudely. This works because the
> filled shape does not have an outline to it. (Meaning the errors in it
> are hidden by the true outline that is created from arcs) As setting
> line width to 0 does not result in a polygon of 0 width the trick with
> the large negative value was used to achieve this.

I understand.
But a better way could be using the minimal width (1 mil) because 0 is
(for historical reasons) selecting the default pen size value.
1 mil should work fine in this case.

A negative value has serious issues:
- As a width it has currently no sense.
- polylines with line thickness <= 0 cannot be selected: it means you
cannot edit/move or delete them. This is the most annoying issue.

Jean-Pierre CHARRAS

Follow ups