On Jun 3, 2008, at 10:24 AM, Peter Clifton wrote:

> On Tue, 2008-06-03 at 08:41 -0600, John Doty wrote:
>> On Jun 3, 2008, at 4:22 AM, Peter Clifton wrote:
>>
>>> On Mon, 2008-06-02 at 19:14 -0600, John Doty wrote:
>>>> On Jun 2, 2008, at 11:02 AM, Peter Clifton wrote:
>>>
>>>> What is "make" supposed to do with a dialog box?
>>>
>>> Nothing more than when it is expected to design your circuits for  
>>> you.
>>
>> I expect "make" to do what it does for software: propagate changes as
>> needed. If I change the package of the default resistor in a project
>> and type "make", I expect all data products for that project to
>> reflect the change.
>
> Ok, then you'd either have it invoke the appropriate command to update
> all elements from the command line (within the Makefile), or perhaps
> some config would be set to make the tools always prefer the library

AAARGHH! NOT THE *LIBRARY*! THE *PROJECT* *SYMBOL* *REPOSITORY*. Of  
course, we don't currently make a clear distinction here, and that's  
the root cause of the problems that embedding and gattrib are  
inadequate at solving.

> copy of a resource rather than the cached copy (just warn in the  
> GUI if
> it detects a change).
>
> I thought you were arguing in favour of the previous behaviour,  
> where it
> is pretty hard, if impossible to auto-update the footprints in a .pcb
> file.

That's because you don't use a project-specific copy of the symbol.  
If you do, it's just Hs, change the footprint, make.

>
>> Unacceptable. In a large project, such warnings are lost in the spew.
>
> You can't have it both ways.. a GUI dialog which warns you something
> broke, or something which can work from the command-line.
>
> What we should do is reduce un-necessary stdout / stderr spew, so that
> any output which is printed is important information for you to  
> review.

A worthy goal. Never achieved in any large-scale use of make that  
I've ever seen, though.

>
>> When I change a symbol to fix a pin assignment, part number,
>> footprint, underlying hierarchical schematic, etc., I expect that
>> change to propagate. Among other things, it's part of design reuse:
>> in prototype maybe the default resistor is 0603, but in production
>> it's 0201. Some customers imagine hermetic packages improve
>> reliability, others are more sensible.
>
> PCB's behaviour is the complete opposite of this, no automatic updates
> are possible.

I don't use PCB. My customers all have their own PCB flows. But here,  
the issue for me is schematic reuse: the layout generally has to be  
redone for any new project, even if most of the schematics are reused  
from previous ones.

>
> For either case, without caching some hash at least (if not a complete
> copy, like PCB), there is no hope for ANY user to be notified about
> library changes. If you don't want notifications / to use the cached
> copy, there is no reason I can see why it couldn't be turned off.
>
>> With symbol collections customized for the project, it's easy to move
>> schematics from project to project (especially if you turn off
>> automatic promotion of device and footprint attributes). So,
>> low_noise_npn.sym represents BCX70K in an SOT23 in one project,
>> 2N930JANTXV in a can in another.
>>
>>>
>>> In the case of divergent footprints / symbols which the user  
>>> _wants_,
>>> I'd imagine the GUI action to express this would then take the  
>>> desired
>>> symbol and dissociate it from the library copy, either giving it a
>>> different name, or just flagging so in the file.
>>
>> It seems to me that *all* project symbols should be disassociated
>> from the library copy and placed in the project symbol repository
>> (often under another name: low_noise_npn, small_resistor, etc.). But
>> embedding them in a schematic is a severe barrier to schematic reuse
>> in an different project.
>
> I was talking about caching of a copy, or a hash value, not full
> embedding with no subsequent use of external files.
>
> (PS. My current project only has 40 non-library symbols, and
> coincidentally, about 40 schematic pages after Makefile + sed based
> auto-generation of a few multi-channel sections).
>
> -- 
> Peter Clifton
>
> Electrical Engineering Division,
> Engineering Department,
> University of Cambridge,
> 9, JJ Thomson Avenue,
> Cambridge
> CB3 0FA
>
> Tel: +44 (0)7729 980173 - (No signal in the lab!)
>
>
>
> _______________________________________________
> geda-dev mailing list
> [email protected]
> http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev

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