On Tue, Dec 2, 2008 at 12:15 AM, al davis <[EMAIL PROTECTED]> wrote:
>
> All I'm suggesting is using a netlist format, and augmenting it
> with info relevant to schematic and layout, while keeping it
> strictly as a netlist format, so you can do a round trip.

That would be very useful but I don't think a common netlist format is
the best tool for this job. Commercial vendors go rather in the
direction of a common design database. Whether it will work for them
or not - it's another story. These same vendors are well known for
locking down customers and competitors.

Note that database concept is not about storage (text backend is
perfectly fine for most applications) - it's about API.

There is a couple of reasons why common API could be better than a
common netlist format:
- API is (or can be made) abstract, netlist format is not.That's huge
benefit for maintenance. It is much easier to add new features at API
level than at raw data level.
- API usually comes with an implementation. For a file format a user
still needs to write functions to traverse data structures etc. Sure,
they can be abstracted out (think DOM or SAX), but this effectively is
building an ad-hoc API on top of data.
- gEDA, thanks to libgeda, is not too far away. It shouldn't be very
difficult for gnucap to link to libgeda and read design data directly
directly from the "database". Sure, libgeda would have to be extended
a bit for this to be possible.
- API can cross process boundaries. This makes tighter integration of
tools (like selecting probe points for gnucap or backannotating
simulated OP results on schematics) possible to implement.

On the other hand, common data format wins in network environment -
you can send a netlist to someone but you can't send API (unless you
send the whole program). For me, a perfect solution would be a common
data format (not necessarily a netlist) for exchanging design data and
an API/database to work with them.

Regards,
-r.


_______________________________________________
geda-dev mailing list
[email protected]
http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev

Reply via email to