kicad-developers team mailing list archive

[Wayne Stambaugh <stambaughw@xxxxxxxxx>]

On 8/13/2015 12:38 PM, Lorenzo Marcantonio wrote:
> On Thu, 13 Aug 2015 14:59:49 +0200,
> David Novak wrote:
> 
>> That's exactly what I told the contract manufacturer, but they insisted
>> on a change. With a shape for every reference designator, they had to
>> review the shape for every part. With this change, they only have to
>> review each shape type.
> 
> They actually use the gencad file for something other than pad
> probing... and even that usage is made obsolete by D356...
> 
>> The change detects when a part is different from the library and creates
>> new shapes in the GenCAD file as needed.

Wouldn't you also want to test the footprints on the board as well?
Let's say you have multiple "smt_resistor_lib:r0805" footprints on your
board and you modified more than one of them with same changes but
different from the original footprint imported from the library.  You
would still have multiple identical footprints exported to the gencad
file which your contract manufacturer will have to review.

> 
> Cool:P You had to do a module comparation function to handle that. Not
> quite easy to pull (tried that, failed and then you have the current
> gencad exporter:D)
> 
>> We have some footprints where multiple pins are left un-numbered. These
>> pins were represented in the GenCAD file as "none" and the multiple
>> "none" pin numbers were causing problems. To prevent the problem, we
>> added an enumerated suffix to duplicate pin numbers.
> 
> That would be the correct gencad behaviour, fully agree with that. Could
> be useful even for named pins which are actually always tied together
> (like ground lugs in shielding can).
> 
>> I agree completely. Our changes are based on the GenCAD standard.
> 
> I hope that gencad die one day or another, like 274D (the external
> aperture gerber format). Too bad that there is no agreed substitute;
> ODB++ is not free, gencam and IPC350 are almost unheard outside their
> designers... maybe because gencam with its 40+ XML schemas is a serious
> contender with STEP on the horribly complex kitchen sink file format:P
> 
> IPC350 seems sane (the downstream CAM offer substantially uses that
> format in core) but AFAIK not established enough to warrant the effort
> to code an exporter.
> 
> The saddest thing is when the fabricator asks for the "MAX file" (*old*
> orcad layout) or "eagle file"...
>