On Jul 11, 2007, at 11:47 AM, Peter TB Brett wrote:

> On Wednesday 11 July 2007 18:10:22 John Doty wrote:
>> On Jul 11, 2007, at 5:38 AM, Ales Hvezda wrote:
>>> Hi All,
>>>
>>> Question for the list, how would you feel if gschem's default  
>>> storage
>>> mechanism changed from referencing to embedding placed symbols?
>>> Instead of
>>> just referencing the symbols, the symbols would be automatically
>>> embedded
>>> into the schematic.
>>
>> Given that you can't edit an embedded symbol right now, I think this
>> undesirable. Especially since I often find myself making wholesale
>> alterations to a multi-sheet design by editing a symbol. AD5601 not
>> available? Easy to switch to AD5621 when your symbols are in a
>> project directory, harder when they're embedded. And in ASIC design,
>> I'm often tinkering with the building blocks. There, a stale symbol
>> could be big trouble.
>>
>> I'm notorious for getting pin numbers wrong. Easy to fix with
>> referenced symbols, not so easy with embedded.
>>
>
> This is from my idea of only having a single copy of the symbol  
> embedded which
> is referenced by one or more instances.
>
> When you update one instance, you update them all.
>

What if your design has many pages? In my AD5601 example, there were  
many of these spread over a number of pages. What about projects that  
contain several boards built from a common parts stock? What about  
subsystems shared between projects? Projects fork and/or merge. They  
contain things other than circuits: software, test data,  
documentation, ...

Given that a modern computer system comes with very flexible and  
powerful mechanisms to manage files, I find it extremely convenient  
to "factor" projects into small files. Manage them with a revision  
control system. Build products (netlists, docs, microcode, etc.) with  
Makefiles. Refactor, fork, and merge largely by moving and  
customizing files.

Perhaps for a single project on a single page, the embedded approach  
is better. But I have a web of related projects, some very complex.  
That gEDA's extreme flexibility makes this manageable is a strong  
reason to use gEDA. Please don't compromise that flexibility.

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
[EMAIL PROTECTED]




_______________________________________________
geda-dev mailing list
[email protected]
http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev

Reply via email to