kicad-developers team mailing list archive

[Eeli Kaikkonen <eeli.kaikkonen@xxxxxxxxx>]

2018-03-21 20:16 GMT+02:00 Wayne Stambaugh <stambaughw@xxxxxxxxx>:

> Sounds reasonable.  If I'm reading the comments correctly, I don't like
> the idea that the current proposed patch leaves footprints in a state
> where they cannot be selected and edited.  I'm sure that will quickly
> produce another but report.
>
>
Now I have to say again that the fix doesn't actually cause that situation.
It happens already. If there is a footprint with only pads with e.g. paste
layer only and then the paste layer is removed from the board, the
footprint can't be selected. It happens already without the patch.
Apparently my bug report was the first one and the patch wouldn't trigger
more problems. Of course I understand that footprints should be always
selectable, and the fact they are not is a bug on its own.

BTW, also the situation when such a footprint is added to a board which
lacks the necessary layer should be handled. It also leads to a
non-selectable footprint.

But now, let's be realistic. What are the odds that people have footprints
which don't have copper pads but have paste-only pads or mask-only pads and
then they remove that layer, leaving the footprint non-selectable? After
all, this happens *only* when a pad has only paste or mask and no copper
*and* the footprint doesn't have any pads with other layers. I have seen
enough to have learned to never say "why would anybody ever do that", but
it seems very unlikely, and removing a copper layer which already has pads
seems also very unlikely (well... it seems to be impossible to do it from
the UI).

Another thing to remember is that my original report was about having data
lost by accident because there was no warning. If users are warned about
consequences when they do something they do it at their own risk, right? I
won't run into the same situation again even if the bug remains unfixed, so
I don't have an axe to grind, but it would feel weird if it wouldn't be
fixed now, considering that even the patch wouldn't lead to any *new*
problems and that the original problem would have been prevented by only
adding necessary checks to the code and a warning to the dialog.

Eeli Kaikkonen