Postscript?!?! (was ASM-1 homepage improvement!)

[Magnus Danielson]

>>>>> "DT" == Don Tillman <don at till.com> writes:

 DT>    From: Magnus Danielson <magnus at analogue.org>
 DT>    Date: Sun, 22 Nov 1998 05:54:44 +0100

 DT>    I just added the ADSR schematic to the ASM-1 homepage.
 DT> [...]
 DT>    (building a Postscript schematic the way I do it requires many
 DT>    renderings of the Postscript, one for each try). 

 DT> Postscript?!?!  What?!?!

 DT> <checking the ASM-1 page...>  Postscript/Emacs/Ghostscript/PDF!?!?!
 DT> What a crazy way to draw a schematic!  This is very cool.  I like it.

Well, since I am at least a little bit crazy or at least a fairly
obscure person, I tend to have habits that follow along with such a
personallity ;)

 DT> The nice thing about this approach is the type of control you have
 DT> over the schematic -- "I want a transistor to look like *this*".  Then
 DT> later you can say, "no wait a second, it'd be better if a transistor
 DT> looked like *this*" and all the transistors are updated.  If you have
 DT> schematic objects dependent upon other objects (ie., postscript
 DT> subroutines) they get updated too.  Schematic objects can draw
 DT> dependant upon given parameters.  Positioning can be done correctly
 DT> too; "place this resistor 2.4 inches north of the collector terminal
 DT> of that transistor", that sort of thing.

Well, not only that. If you would actually read my postscript code
(yeap, there is even comments in there) you would find out that I have
routines so that I can say "I want a transistor here with it's
collector pointing left-wards". The associated information like Q12,
2N3055 is also being placed as appropriately depending on how the
symbol is being mirrored and rotated. If I need a transconductance
op-amp symbol I can take the source of the standard op-amp add a new
input signal and a few rings, adjust the associated code and I am free
to use it (see the result in the ASM-1 VCF). There is ofcourse cases
when I cheet (see the ASM-1 VCO where the LM311 is so sparse that it
does not tell the full story of the function in the cursuit, a
definite must-fix).

The system that I have built will only require a few fix-points, the
rest is relative positions in long sequences.

 DT> I've done some PC board designs in a similiar way (but with a
 DT> Lisp-based drawing program in between; it'd probably have been a lot
 DT> easier to go directly to Postscipt).

Maybe. I once did a PCB layout thing in AutoCAD using AutoLISP, it
could support some 250 layers but the main reason it would not work
for such a thing is just the sheer size of the project and no netlist
support would be a large problem, not the direct support for that many
layers.

The downside to this form of editing is that it is not very well
suited, except that it does not require any commercial software.
Actually, it is quite painfull. However, the experience I have with
other schematic systems is that those are also painfull. Also, you
should not expect a netlist being generated from this system...
Maybe I should make a protecting language (just like TeX) ;)

Cheers,
Magnus