Re: gEDA-user: Resistor values…

[John Doty <jpd@xxxxxxxxx>]

On Dec 26, 2010, at 4:14 AM, Vanessa Ezekowitz wrote:

> On Sat, 25 Dec 2010 12:44:54 -0500
> John Doty <jpd@xxxxxxxxx> wrote:
> 
>>> "Often", perhaps, but not usually.  No matter how you slice it, the most
>>> common way to use such a symbol and its corresponding physical
>>> representation is as a component on a circuit board or in an IC.
>> 
>> Maybe for you. But gEDA isn't limited to that kind of flow. I pray that it
>> will remain flexible, not specific to any particular kind of flow.
> 
> Flexibility and specific applicability are not mutually exclusive, and for the very reasons you are citing here.

True, but what makes this possible? It's *avoiding* specificity in the foundations.

> 
> If your setup is such that these "value=" attributes cause problems, I submit that the problem is not in the presence of these attributes, but in your workflow.  You've become accustomed to a certain pattern and have not planned far enough ahead to account for relatively minor changes.

I used the existing library symbols as they were. The fundamental logic of the library design in gEDA is that semantics should not change, because that can break existing designs. But it's easy to make specialized libraries, so that's the solution.

I do not call putting in an attribute that triggers special processing in many back ends a "relatively minor change".

> 
>>> The existence of something does not imply the requirement to use it.
>>> 
>>>> gschem/gnetlist are excellent tools for
>>>> schematic capture for VLSI, symbolic analysis, and simulation.
>>> 
>>> And they will continue to be.
>> 
>> They won't if the attitude of "I don't care to know about any flow except
>> pcb, and all I want is my version of the pcb flow" isn't vigorously opposed.
> 
> Nor will they be if you continue with your attitude of "I don't care about any workflow other than VLSI simulation".

What??? Of course I care about a wide variety of flows, including (several) printed circuit flows. I have contributed two printed circuit back ends for gnetlist to the gEDA project. 

> 
> The project needs to be able to cater to both as well as it can, even if that means the users need to make minor changes to the way they use it.

Fixing problems in a library of schematics going back eight years isn't "minor". I frequently reuse schematics. When reusing a symbolic analysis schematic, an unnoticed "value=0" attribute could cause serious problems.

> 
> You confuse a request for a feature (or in this case, a request for a modified default) with the desire to eliminate some feature you find critical.

Once the default is in use, it *is* a critical feature to those who have used it. Changing it breaks things.

> And your point is?  You're writing formulas with refdes's, not value= entries.

Depends. If I want to set a constant value in the schematic rather than leaving it variable, value= fixes it. Does that not make sense?

> 
> No matter how complicated your resistor formula are - sooner or later, those equations HAVE to boil down to some number of ohms or your circuit isn't going to do what it should.

But that happens later. A schematic for symbolic analysis is like a textbook abstraction: idealized components of infinite dynamic range with power coming out of nowhere. The resulting equations drive component value selection in a more realistic schematic downstream.

>> 
>> A nice simple text editor is a good thing. Versatile. A WYSIWYG word
>> processor is much less versatile.
> 
> How so?  Name me one thing a simple text editor can do that a properly-written WYSIWYG word processor supposedly cannot.

Not waste your time with endless point and click. Clearly separate content from markup, thereby putting the user in control.

> 
>> Ah, the tyranny of the majority again.
>> 
>>> All we are proposing here is adding some reasonable, sane defaults for
>>> things like "value" - things you can ignore if they aren't useful to your
>>> particular work.
>> 
>> If you add a value= attribute to resistor-1.sym it will break most of my
>> symbolic analysis schematics because of the inheritance rules. The presence
>> of specific values overrides the default of "use the refdes symbolically".
> 
> If you don't set up your work environment so that it can grow along with and adapt to changes in the tools that it uses, you are just begging for trouble.

The way to grow gEDA is by adding optional modules: new libraries, back ends, scripts. Changes to the core should maintain compatibility with existing designs. Additions to the core API could facilitate new kinds of add-ons.

>  If your rules get broken by simply adding an attribute that is normally absent, then you need to fix your setup to not depend on it, rather than trying to force the software to leave out the offending attribute.

Who said anything about forcing the software to leave it out? The software should put it in if it's in the symbol, leave it out if it's not. If the symbol is wrong (and most are no doubt wrong in the context of any specific flow), create a *new* symbol. Every existing symbol was once considered good for its purpose, whatever that was, and you cannot know what a change might break.

> 
>> Again, the right answer is not to change the default library, but to create
>> libraries for specific purposes on gedasymbols.
> 
> Then I submit that VLSI-specific symbols should be created.

For a start:

http://www.gedasymbols.org/user/john_doty/symbols/openIP/

>  Let the existing symbol library remain generic as it is now, just build up the contents of the individual symbols a little bit - a description here, a datasheet reference there, a default value, even a default footprint where it makes sense.

No. Leave the existing library alone other than fixing broken links or descriptions. Changes to critical attributes will cause much grief. And it's just as easy to copy the symbols to a new library (using new names) and fix them up to meet whatever spec you imagine will be better. Be prepared for disagreement on what "better" means from other users (but I'll cheer a new library).

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