kicad-developers team mailing list archive

[Wayne Stambaugh <stambaughw@xxxxxxxxx>]

On 4/12/2016 2:34 AM, Lorenzo Marcantonio wrote:
> On Tue, Apr 12, 2016 at 07:58:05AM +1000, Cirilo Bernardo wrote:
> 
>> a) ability to use "." or "," regardless of the language setting - assuming
>> no sensible person ever enters a thousands separator in CAD software
>> (I don't know any CAD software which accepts thousands separators).
> 
> Already doing this :D
> 
>> b) ability to specify a unit as a suffix:  0.001m, 1mm, 0.03937in,
>> also 'mil' and for the Australians 'thou'.
> 
> Already doind that, too :D
> 
> IIRC it's a couple of function, not a validator-thingie, it was
> literally years ago. But I suppose it's easily convertible.
> 
>> People have also been asking to have angle suffixes: rad/deg/mrad
> 
> Seriously... *what use* are radians in a pcb layout? The only thing that
> comes to mind are arc stub for impedance match...
> 
>> but in my opinion such a thing would have to be implemented in
>> a different validator - maybe an AngleFloatValidator vs a
>> UnitFloatValidator  - or else the validator will need to be able to
>> change its behavior based on some kind of flag.
> 
> Please please no flag. Two different functions/classes/whatever.

+1.  I am a firm supporter of the KISS principal.

> 
>> The current input text boxes in my opinion have a number of bugs:
>> 1. you can put alphabetic characters anywhere
>> 2. if the unit is not known or a mistake is made then the conversion
>> result is "0" which is obviously wrong. In such a case the validator
>> should prevent the user from changing the value at all and somehow
>> signal the user of the mistake so the user can correct it. Naturally it
>> is impossible to tell if there is a mistake with "m" vs "mm".
> 
> OK, from memory, the current value conversion (as you said it's *not* a
> validation) occurs at data transfer time, i.e. from variable to control
> and back. It's a conversion since the variables are in internal units
> (mils in eeschema and 10 nanos in pcbnew). That's probably one of the
> reason for the tenth-degree input, there is no *conversion* pair for
> them.
> 
> If the idea is to move these feature in a while-type validation, I think
> it's good (but be careful at the 'intermediate' stages while people is
> editing the content).
> 
> 
> 
> _______________________________________________
> Mailing list: https://launchpad.net/~kicad-developers
> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~kicad-developers
> More help   : https://help.launchpad.net/ListHelp
>