kicad-developers team mailing list archive

[Dick Hollenbeck <dick@xxxxxxxxxxx>]

On 04/11/2013 09:11 AM, Lorenzo Marcantonio wrote:
> On Thu, Apr 11, 2013 at 08:57:24AM -0500, Dick Hollenbeck wrote:
> 
>> If I can broaden this topic somewhat, this type of on the fly board
>> change (during loading) falls into a general category of "doctoring a
>> board during loading".   Let's call it "board doctoring".
>>
>> Generally it makes me nervous, even though it may have its place.
> 
> Personally I'd have put a warning to it, instead of silently eating the
> changes. Unless it was for something which can't actually work (like PCB
> edges in modules, if it's still the case). On such a thing I'd either:
> 
> a) refuse the board (like a missing parameter or something)
> 
> or
> 
> b) notify during loading *and* forcing it
> 
> In the specific case a good solution would be allow the text on the
> copper layers but maybe emit a warning in the DRC log (something like
> 'warning text on copper will not be checked blah blah blah')
> 
>> I suggest that we raise the threshold under which we would do board
>> doctoring to a higher level of criteria.
> 
> What do you mean?
> 
>> For low value fixes it is best simply to ask the user to hand edit his
>> board file.
> 
> Do you mean 'that layer is not good, fix it by hand' like in my a)
> scenario?
> 
>> a) use hand edited boards to gainfully do an end run around a weak
>> user interface and achieve a more full featured board manufacturing.
> 
> Agree on that, it's not desiderable to need hand editing but in many
> case it's convenient.
> 
>> b) track down editing bugs, because they can be masked behind board
>> doctoring.
> 
> Sorry but I can't see an example for this...
> 
>> Board doctoring is a practice we should frown upon, and only use
>> reluctantly.
> 
> IMHO it shouldn't needed *if* the file is correct. An exception could be
> loading from a previous version where some semantic changed (like text
> style in eeschema when wasn't yet introduced)
> 
> I think that kind of fix is from the times when (probably) modules were
> allowed on silk and adhesive layers (this is not possible these day).
> 


I suspect that Jean-Pierre and Wayne will be able to make more sense
of my posting than you have.