[Gnucap-devel] Re: gEDA: ACS is now gnucap .. new upload.

[Al Davis]

I am forwarding the reply to the gnucap develop list, too.

On Thursday 08 November 2001 10:57 am, Uwe Bonnes wrote:

> Either make a new gprint (or similar named ) command  that allows
> gnuplot to take the print output unchanged or change the print
> output. 

One of these days I hope to be able to feed gnuplot the way octave 
does.  Any ideas of how to do this?

The existing "plot" is designed to drive an external plot program, 
but the protocol was designed 20 years ago, and is needs updating.

> ...... There are two problems I see with the present print:
> - The Prologue "GnuCap 0.30\nThe Gnu Circuit Analysis Package..."
> isn't prefixed as comment and needs to be edited manually.

When I run a batch regression suite, using the whole output, I remove 
the headers with "tail -10".  I suppose it could be prefixed as a 
comment.  There are other commands (list, status, fanout, ....) that 
produce output that doesn't fit gnuplot and other plotting programs, 
so I am not sure that it would be of much benefit.  It is trivial to 
change it.  Comments????

Also ....  You can direct the output of a particular command to a 
file with ">".  I need to add "|" too.

Questions:

Should ">" be allowed to clobber?  At present, it does not clobber 
without permission.  Proposed change:  option clobber/noclobber.  
Which should be the default?

> - Gnuplot isn't happy with like numbers like "500.p" or 2.5K (yes,
> it is high voltage). Numbers need to be in exponential form to be
> readable by gnuplot.
>
> If there is a trick to circumvent this problem, let me know.
>

If you redirect the individual commands, or if the program knows it 
is going to a file or pipe, it puts it in standard scientific 
notation.  You can force it with the option "basic" on any command.

Questions:

Which way should be the default?  Here is another place for a global 
option.  Which should be the default?  I like it the way it is, but 
agree the global option is needed.

> Btw: Is there an easy way to variate a device parameter, like the
> inductance value and print out the result. This is good to see the
> sensivity of circuit. As a workaround, I duplicated the circuit
> with different node numbers and different values of the inductance
> and printed the values of the interesting nodes. A definition like
> L1 2 3 (0 1u 0.1u)
> would come handy :-)

There is a "sweep" command that lets you do this.  One item on the 
"todo" list is to make it more general, and more like traditional 
programming.  As it stands, it can only change plain values, not 
named ones, and you cannot nest loops.  After the sweep is done, the 
values are restored.

The command "modify" changes a value permanently.  The command 
"fault" changed it temporarily, so you can change it back with 
"unfault".

al.