Re: gEDA-dev: Thoughts about the PCB database model

["Igor Izyumin" <igor_dot_izyumin-at-NOSPAMgmail_dot_com>]

On 6/28/07, DJ Delorie <dj@delorie.com> wrote:
>
> The most likely change I've made is to fiddle with the pin/pad copper
> sizes, or clearances or mask openings.

I suppose this can be a time-saving feature when dealing with small
boards and library footprints, so it should probably be done.  I think
it could be implemented just by copying the footprint master with the
modifications.  If the user wants to update the footprints, they could
leave the copy alone or replace it with the original.  Eventually,
some sort of merge algorithm could try to resolve conflicts.

> Also, the location of the
> refdes is part of the element, and we move that around a lot, just as
> we change the refdes text and other texts.

A refdes should not actually be part of the footprint master, it's an
attribute that gets attached to each instance.  Right now, it's saved
as simply an XY coordinate offset from the part's centroid.  This area
needs improvement, anyway -- for instance, it's impossible to show the
refdes and the value at the same time, since there is only one text
location that gets saved.  I would implement it as an attribute that
gets attached to an instance when the instance is created, which can
get rendered into a physical text object if its display is enabled.
A more flexible implementation could allow all sorts of nice features,
like showing part numbers or tolerances or assembly notes imported
from the schematic.

> But still, we need to avoid excess computations in the redraw path,
> because redraw is already fairly busy.  A cache is fine, or keep an
> "unrotated" element somewhere, or whatnot.

The redraw path should definitely be kept out of this.  Having a cache
also helps a lot with online DRC and that kind of thing.


_______________________________________________
geda-dev mailing list
geda-dev@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev