kicad-developers team mailing list archive

[inkblotter <inkbottle@xxxxxxxxx>]

On 03/13/2015 04:00 PM, Wayne Stambaugh wrote:
> I finally had a chance to test this.  Here are my concerns:
>
> * The Ctrl+mouse wheel behavior is completely broken on windows and I
> suspect Linux as well when the use mouse wheel to pan option is set.  It
> acts just the Shift+mouse scroll wheel (scroll vertically).  This needs
> to be fixed.  The default behavior should always remain the same across
> platforms.  It can be OSX specific compiled code or a configurable
> option but breaking the current behavior on other platforms is not an
> option.
>
> * The Ctrl+wheel mouse when the use mouse to pan option is not set
> (zoom) periodically freezes.  I seems to come back when I perform some
> other function that updates the display but the exact cause I cannot say
> for sure.
>
> I did not get a chance to try it out on a laptop to test the track pad
> behavior.  It definitely needs some work before it can be committed to
> the product branch.
>
> On 3/4/2015 7:31 AM, Bob Gustafson wrote:
> > My last build last week seemed to work fine with both up-down and
> > left-right panning using the trackpad.
> >
> > If you want to support OSX in the stable release, I would say that
> > trackpad panning is important.
> >
> > I won't be able to do much testing until after next Tuesday.
> >
> > Bob G


Ah yes, reality hits the fan, as it often does.

At least on Fedora 21, I can pan with the mouse wheel depressed and it 
correctly pans the display for all intents and purposes. And if I roll 
the mouse wheel  it zooms.


Everyone is practicing how to build  Kicad from src, because after the 
stable  release you/we plan to remove all that spaghetti cmake stuff 
that we all love to hate about cmake.   I, for instance, have taken to 
building release and debug versions of Kicad every now and then with all 
their Kicad dependencies built and you can verify that with ldd.   I 
have largely succeeded despite lots of pointless complains from cmake.   
That is, I have built

boost 1.5.7
wxWidgets 3.0.2.0
pixman  git head version
gl enabled cairo git head, essentially 10.2

Bleeding edge all. Kicad hasn't blinked.  Kicad doesn't really care 
about versions; it cares only that certain calls exist in whichever 
version of dependencies it has.

I also have installed and built wxPython_Phoenix,  built via pip, but 
not in the way kicad approves of.  There should be a patch in wxPython 
head to make wxPython (any version) build using the picky gcc 4.9.2 
compiler.  But the wxPython people have since run off again and are 
working on wxPython_Phoenix now.

When I get wxPython correctly built against wxWidgets I would try to 
figure out how to make Kicad  build with the
-DKICAD_USE_WEBKIT=ON flag on Fedora 21.

I will also try building all the dependencies, and Kicad, with clang and 
clang++.  If it works on OS X, then it should work on Linux as well, yes?

Ideally, there would be an external (from Kicad) script, jar or python 
program that would attempt to maintain some semblance of constraints 
only for that platform.  cmake would do what it does fairly well: it 
builds the code and  installs it.  It could use some more work on the 
make clean target that really makes clean, instead of the make 
mostlyclean we have now.