X-Authentication-Warning:	delorie.com: mail set sender to geda-user-bounces using -f
X-Recipient:	geda-user AT delorie DOT com
X-Envelope-From:	paubert AT iram DOT es
Date:	Sat, 1 Jun 2013 01:38:24 +0200
From:	Gabriel Paubert <paubert AT iram DOT es>
To:	geda-user AT delorie DOT com
Subject:	Re: [geda-user] [RFC][PATCH] Separate aperture and drill number
	spaces.
Message-ID:	<20130531233824.GA19544@visitor2.iram.es>
References:	<20130531221316 DOT GA14164 AT visitor2 DOT iram DOT es>
	<201305312228 DOT r4VMSBpu002091 AT envy DOT delorie DOT com>
MIME-Version:	1.0
In-Reply-To:	<201305312228.r4VMSBpu002091@envy.delorie.com>
User-Agent:	Mutt/1.5.20 (2009-06-14)
X-Spamina-Bogosity:	Unsure
X-Spam-Score:	1.3 (+)
X-Spam-Report:	Content analysis details: (1.3 points)
	pts rule name description
	---- ---------------------- --------------------------------------------------
	-1.4 ALL_TRUSTED Passed through trusted hosts only via SMTP
	2.7 TRACKER_ID BODY: Incorporates a tracking ID number
Reply-To:	geda-user AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	geda-user AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On Fri, May 31, 2013 at 06:28:11PM -0400, DJ Delorie wrote:
> 
> I suggest checking through the archives and seeing if there was a
> reason we did it that way.  We've seen a *lot* of fabs have strange
> expectations, so any change "because it seems right" is automatically
> suspect ;-)

I would not be surprised if some manufacturers did wrong things with
Gerber files for the same PCB having different definitions for the
same aperture on different layers. But between Gerber and drill files, 
which are not even processed by the same machines, this looks very strange. 

In any case, it looks like this was introduced by commit 

44c1a1cf0463ae311464144a6a6ae88e0d626e13

and the commit comment is "use a global aperture list instead of a
per layer list". I believe that the fact that this also affects
drill files is a side effect or collateral damage.

[I did not compile the before and after version of the code, but I
think that it is where it basically started.]

BTW, there were many changes after that patch, and now we have independent
non overlapping aperture lists for all files in a project, this makes
aperture numbers explode and they might exceed the 999 limit of earlier
Gerber specifications (it has been raised to INT_MAX=~2×10⁹ only recently):
my test boards reaches aperture 202 for a 6 layer boards with a limited 
variety of component pads and line widths.  In addition gerbv tells me 
that several aperture definitions for each layer are unused, hint of 
another bug wasting aperture space.

Everytime you add a layer, the current scheme increases the number of 
apertures (by at least the number of different line widths used on that 
layer plus the number of via sizes and through hole pin shape/size
combinations).

I'm seriously thinking of adding an option for aperture allocation:
1) non-overlapping: like it is now
2) global: a single global number space (reuse numbers for 
   identical apertures between layers)
3) independent: start at D11 (or D10 actually) for each layer

I think that 2) above should be the default.

	Gabriel

- Raw text -