kicad-developers team mailing list archive

[José Ignacio <jose.cyborg@xxxxxxxxx>]

A subtle bug still remains unfortunately, if you try to move the same
selection multiple times using different anchor points you can't because
once the reference point is set by updateModificationPoint, it is never
unset until the selection is destroyed. perhaps it should be unset whenever
a transfomation is done.

On Sun, Nov 19, 2017 at 8:17 PM, José Ignacio <jose.cyborg@xxxxxxxxx> wrote:

> I think i found a possible fix that avoids interfering with the copy and
> paste origin logic added (even on single element pastes):
>
> diff --git a/pcbnew/tools/edit_tool.cpp b/pcbnew/tools/edit_tool.cpp
> index bb6fe3f80..f8ff9ce32 100644
> --- a/pcbnew/tools/edit_tool.cpp
> +++ b/pcbnew/tools/edit_tool.cpp
> @@ -450,8 +450,6 @@ int EDIT_TOOL::Main( const TOOL_EVENT& aEvent )
>
>                      m_cursor = controls->GetCursorPosition();
>
> -                    updateModificationPoint( selection );
> -
>
> This call guarantees that selection.HasReferencePoint() always returns
> true, so the other two code paths never get executed, so i removed it, as
> it is not necessary in the first if case:
>
>
>                      if ( selection.HasReferencePoint() )
>                      {
>                          // start moving with the reference point attached
> to the cursor
>
>
> And then added it back for the other two cases:
>
>
> @@ -469,11 +467,13 @@ int EDIT_TOOL::Main( const TOOL_EVENT& aEvent )
>                      {
>                          // Set the current cursor position to the first
> dragged item origin, so the
>                          // movement vector could be computed later
> +                        updateModificationPoint( selection );
>                          m_cursor = grid.BestDragOrigin(
> originalCursorPos, curr_item );
>                          grid.SetAuxAxes( true, m_cursor );
>                      }
>                      else
>                      {
> +                        updateModificationPoint( selection );
>                          m_cursor = grid.Align( m_cursor );
>                      }
>
>
> Patch file attached.
>
>
> Thanks,
> Jose
>
>
> On Sun, Nov 19, 2017 at 6:04 PM, José Ignacio <jose.cyborg@xxxxxxxxx>
> wrote:
>
>> I can confirm this, it also affects the movement of graphic items like
>> lines and arcs to very disturbing results. It will always use the
>> centerpoint of an arc or first point of a line, snapping the part to the
>> cursor every time. I ran a git bisect and it looks like the first commit
>> that introduced that behavior was  57310001350ead3a6b7870f19982f77b88ac5b8a,
>> with the introduction of storage of offsets in the SELECTION class.
>>
>> I couldn't find a satisfactory fix because i'm not familiar with the
>> design of the edit tool, the expected behavior is that when you press M
>> with the cursor over an endpoint of a line or arc, kicad will use that
>> endpoint as the reference anchor to move the object around.
>>
>> Thanks,
>> Jose
>>
>> On Sun, Nov 19, 2017 at 3:05 PM, <d.b2@xxxxxx> wrote:
>>
>>> The changes introduced in #931a1ccaff, that bring new features to
>>> copy/paste in pcbnew, change the behaviour of the cursor snapping when
>>> moving footprints:
>>> Before you could move components by either their origin or any pad. This
>>> is a really handy feature that allows the easy alignment of differently
>>> pitched components without the need to constantly change the grid (And I
>>> don't seem to be the only one who enjoyed this feature:
>>> https://electronics.stackexchange.com/a/214072).
>>>
>>>
>>> Changing line 455 of pcbnew/tools/edit_tool.cpp from
>>>         > if ( selection.HasReferencePoint() )
>>> to
>>>         > if ( false && selection.HasReferencePoint() )
>>>
>>> brings back the old behaviour (but breaks of course other things as
>>> that's just a quick hack). So I prosume it would be quite easy to re-enable
>>> this behaviour, maybe by adding an additional check prior to the query of
>>> selection.HasReferencePoint() to see if the cursor is currently above a
>>> movable pad.
>>>
>>>
>>>
>>> Looking forward to discuss this issue with you guys :)
>>>
>>> Best regards
>>> Daniel
>>>
>>> _______________________________________________
>>> 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
>>>
>>
>>
>