[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: gEDA-user: Collaborative Development of Boards

Hash: SHA1

Am 21.01.2011 01:33, schrieb John Griessen:
> On 01/19/2011 06:23 PM, Markus Hitter wrote:
>> One possible drawback for both ideas: you can't route tracks through
>> the "foreign" area/sub-layout, even if there's enough room after
>> assembling the zones.
> In chip layout, where you do have layout sub-cells definable by the tools,
> all you do for for the route through tracks is put them in the sub-cell
> as a floating unconnected trace that you do LVS on only at a higher level
> of completeness -- when it's with the surroundings.  Floating tracks
> might trigger a DRC, but I think they are perfectly valid and
> I'd rewrite the DRC.  I can't remember if DRC2 or anything else
> complains about floating tracks...
> John


it's interesting that you give an example for collaboration that's in
chip design. Chip design is closer to my (professional) home turf than
board design (just a hobbyist :-) ), but I always saw the limitation
about what I can do in my basement (and limited myself somehow). The
chip design floe I know reserves the lower level metal layers to
sub-cells and the higher level metal layers to "global wiring" in
reserved to  connect the sub-cells (with some simplification).


on a board level collaboration I see basically two different approaches:
1. time slicing
2. area slicing

1. time slicing
One person owns the board for a given period of time, the workflow is:
checkout -- work on the board -- chaeckin and the next person takes
over. This is the approach to use if the contributors are in different
time zones and it really requires godd communication. I think this is
supported by geda out of the box as it boils down to a communication

2. area slicing
This is far more challenging than the work flow described above  The
design needs to be partioned into sub-cells, process them independently,
and do the connections between thte sub-cells on reserved layers. There
are some requirements that the gaf design flow can't fulfill (yet).

Net: the question is how you define "collaboration" that defines your
infrastructure sequential or concurrent updates to the desing library.
- From my experience the time slicing approach is easier to handl and
better supported by tools (cvs, svn...)

- -- 

Mit freundlichen Gruessen

Dietmar Schmunkamp
Version: GnuPG v2.0.15 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/


geda-user mailing list