On Sun, Aug 05, 2012 at 12:11:04AM +0200, Miguel Angel Ajo Pelayo wrote:
Nice proof of concept :-), It's really useful already :-)
Consider that now I have to do 4-5 plot runs setting options by hand on the dialog and combining layer with a shell script. I waste about 30 minutes doing fabrication plots for a 4 layer boards. At this stage it would already reduce time to about 5 minutes (you need to do pick and place, drills and gencam too. Also zip the whole for emailing). So, yes, it's already *very* useful for me (it *only* need a gui and a lot of error checking on parameters:P:P)
Let me understand the interpretation of your pctl file, as I imagine, it
reads options, options options, and then you run an output like HGPL= or
PS= or PDF= ,
The interpretation work this way: options simply sets an option;HPGL=,PS=,PDF=,DXF=,GERBER= starts a new plot (closing the previous one) with current plot options (like scaling)
LAYER= actually does a layer in the current plotfile with all the options in effect.
Changing plot-level options between layer could have, uhm, unexpected results XD (altough for most of them unexpected would be only a no-op)
Looking at the code I identified these as plot-level options:
AutoScale,PlotScale,PlotMirror,
PlotLineWidth,AuxOrigin,PSA4Output,
FineScaleAdjustX, FineScaleAdjustY,
FineWidthAdjust These set the plot coordinate system.
Obviously it doesn't make sense to plot
different layers on the same sheet with
different coordinate systems...
(also the plotter classes simply can't
do that at the moment). FineWidthAdjust
actually *is* processed layer-by-layer but
logically it makes no sense...
TrimSilk, PSNegative
These use layering already for their work.
TrimSilk with multiple layers would give
strange results (but it's only for gerbers
and I don't think the multilayer feature
would be useful here). PSNegative fill the
page in black at opening and then paint in
reversed color so the result would be at
least curious. Actually the whole
'negative print' feature is a little strange,
the only reason for it would be using negative
photoresist... anyway if it's there something
probably is using it
HPGLPenNum,HPGLPenSpeed,
HPGLPenDiameter,HPGLPenOverlay
These *should* be layer-level options, but the
plotter classes don't handle pen change after
open, at the moment. Diameter and overlay are
actually processed layer by layer but they
have no sense without the corresponding pen
change command. Since HPGL is mostly unused
these day I'll fix that only if people ask for
it.
Other options which have 'interesting' behaviour would be:
DrillShape Actually is only on PS (so it would be
trivially added to PDF) using scraping. I
think it would be useful for DXF too, at
least (to have fixturing and tooling holes
references without adding them to the drawing
layer), so I'm thinking to add a primitive
to the plotter engines for them: scraping
would be useful for PS and PDF, DXF could use
circles or points (depending on the mark type),
HPGL maybe circles and for gerber maybe single
flashes (would make a plot like gerbv view of
drill tapes)
PlotFrame At the moment is a layer level option but I
think it should be plot-level (replotting the
frame is not interesting...)
ExcludeEdge Is not strictly needed since you can always
plot the pcbedge as a layer, but it's needed
to keep the preference for the standard plot
dialog... I'll think about that when I convert
the standard plot to use the BatchPlot
class.
I'd also like to add options to set the color for reference and value: you could have DXF plots with both of them forced on (using PlotInvisibleText) and controlling the display with the mcad layer system. Could be *very* useful for repair diagrams in my experience.
when you run an output, then all the previous options get cleared? :)
At the moment no, they accumulate in the same option bank (it's easier when writing the control file by hand), but clearing them would be easy. It's only a design decision. But the plotfile open can tweak some option (for example, setting line width to atleast 1), so that would need some (but not a lot) of thinking.
For a final s-expr format it could be something like:
I'd say more something like:
(...
(output
