Re: gEDA-user: A little puzzled about the purpose of gschem

[John Doty <jpd@xxxxxxxxx>]

On Apr 28, 2010, at 12:41 AM, Madhusudan Singh wrote:

>   On Sat, Apr 24, 2010 at 6:40 PM, John Doty <[1]jpd@xxxxxxxxx> wrote:
> 
>   On Apr 24, 2010, at 7:16 PM, Madhusudan Singh wrote:
>>  Failure to correspond to your prejudices is not imperfection.
>> 
>>  Needing an extra 20 minutes after wiring the net, to populate spice
>>  models for each gschem schematic (instead of having a set of
>   default
>>  libraries that do that for common circuit elements) for each
>   circuit,
> 
>     I really don't think you understand how the process works. Can you
>     post a schematic? All you should be doing is attaching model-name=
>     and perhaps file= (although I prefer to include a library once) to
>     active components. Not so hard.
> 
>   Harder than having a proper library that allows me to focus on the
>   circuit design instead of these kinds of clerical tasks that appear to
>   arise from an utter lack of understanding that human beings do not like
>   to do mundane, repetitive tasks that are best suited for the software
>   to take care of.

And gEDA is better at doing the mundane, repetitive tasks than other systems. Once you have your processes set up, a single "make" can generate netlists, BOM, simulation results, printable schematics, typeset documentation...

>   While we are on the subject, just typing up the entire spice netlist
>   from scratch in a single window is not hard (and arguably easier than
>   scattering it all over the place the way things are set up now) either.

I can't imagine doing that for one of my VLSI designs: I could never get it right. But I don't have to.

>   I can get faster results from MacSpice than this artificially
>   convoluted gschem + patchwork + gnetlist workflow.

Will you please show us your work? Apparently you're doing it in some convoluted way that makes it unusually hard. Yes, gEDA will let you make your processes as difficult as you want, but that's *your* choice, not the property of the tool.

>   I can understand that you have some emotions invested in geda for
>   whatever reason, but your statement above made absolutely no sense.
> 
>     What you consider "common circuit elements" are undoubtedly
>     different from what I commonly use. That's how it goes. You have to
>     build your own library, just like I had to when I was using Pspice
>     back in the '90s.
> 
>   Which is precisely the problem. This isn't the 90s. I grew up on BBC
>   microcomputers. Do I feel nostalgic about the things I used to be able
>   to do with those beauties ? Yes. Do I think that the associated
>   workflow with them was superior to the workflow today (even with
>   something as unreliable as Windows) ? Not a chance.

I'm sure Pspice users have the same problems today, only worse, since the component choices are much wider.

> 
>>  compared to spending 0 extra minutes on something like LTSpice or
>>  PSpice is not prejudice. It is 20 minutes of wasted time. Of
>   course, I
>>  have a decided prejudice against wasted time of that sort. So, very
>>  prejudicially, I view it as an imperfection.
> 
>     The real time wasters aren't the setup, but the the repeated manual
>     operations of GUI tools.
> 
>   Thanks for buttressing my argument:
>   1. If GUI tools are the problem, why use gschem at all ?

As I have said many times on this forum, GUI is suitable for interactions between humans and computers when those interactions are inherently graphical. But GUI is not a good way to *automate* processes that the computer can do by itself.

>   2. If repetition is the problem, why the defense of the current
>   workflow that requires repetition of the task of putting in pieces of
>   spice script in different pretty little boxes ?

For a subcircuit schematic you need a box that effectively says "this is a subcircuit with name...". Other than that, you need no boxes, although sometimes they are convenient. So your problem would seem to be that you don't understand the toolkit. Perhaps the documentation needs improving.

Grab the "development project" from http://www.gedasymbols.org/user/john_doty/models/opamp/index.html and try it out. It's pretty easy, I think, although at this level of simplicity you don't really see the full power of a scripted gEDA flow. A few "pretty little boxes" in the schematics, but they were hardly difficult to create. The two in the top level schematic could be combined if I wanted, so there'd just be one per schematic.

For one of my projects, I have a specialized program (not part of gEDA) that creates an elaborate simulation script, typically ~500 kilobytes, mostly PWM() source data. All I have to do is type "make chaintest" and the Makefile generates the simulation netlists from the gEDA schematics, runs the program to create the simulation script, concatenates the generated simulation script and another much simpler fixed script to the top level netlist, invokes ngspice, and when that finishes runs a pipeline of specialized programs to reduce the massive output of ngspice (up to 20 megabytes, depending on parameters) in this particular case to something humanly understandable. You can't set this kind of thing up effectively with a GUI.

> Even MacSpice is better
>   than that.
> 
>> 
>>  You do have an interesting definition of productivity then. But no
>>  matter.
> 
>     Given that I've designed 6000 transistor VLSI chips and 1000
>     component circuit boards with gschem, I think I understand its
>     productivity. You have to use the power of the toolkit, not struggle
>     against it.
> 
>   Or use a better toolkit that takes that needless, wasteful, and
>   professionally irrelevant struggle out of the equation.

If you're struggling, you're not using the tool effectively. Show us your work. We can help you, and when we figure out why you're puzzled maybe we can improve the documentation.

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd@xxxxxxxxx




_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user