kicad-developers team mailing list archive

[Andrew Lutsenko <anlutsenko@xxxxxxxxx>]

Hi Nick,
Oh, my apologies. Gmail decided to chuck Seth's last message into spam for
whatever reason.

Seth,
I did quite a bit of digging and testing on debian8 VM.
Issue is not gcc, it's because of old swig.

This was fixed in swig 3.0.6, see
https://github.com/swig/swig/commit/428b6176df750e41e3db98f029381af0df2a34c3

But debian8 comes with 3.0.2 by default.
%ignore will not help, swig just can not parse templated friend operator.

Can swig on the qa machine be updated? Or better yet can you upgrade to
debian 9? Debian 9 has swig 3.0.10 and compiles this just fine.
Aside from this debian 8 is very old and should be done away with anyway
because of security, old compilers, etc.

Thanks,
Andrew


On Mon, Jul 30, 2018 at 10:38 PM Nick Østergaard <oe.nick@xxxxxxxxx> wrote:

> Hi Andrew
>
> Seth already shared the build log.
>
> tir. 31. jul. 2018 05.29 skrev Andrew Lutsenko <anlutsenko@xxxxxxxxx>:
>
>> I just noticed that my patch was reverted with reference to a build
>> failure with gcc 4.9.2
>>
>> Could you share the error messages too?
>> Also curios which platform uses such ancient compiler. I tried msys2, it
>> only has gcc back to 5.1 and even debian stable uses 6.3.
>>
>> I'm more than willing to put in the work to fix the issue but would like
>> some pointers on how to replicate it, unless it's something I can easily
>> figure out from error listing.
>>
>> Thanks,
>> Andrew
>>
>> On Mon, Jul 30, 2018 at 9:58 AM Andrew Lutsenko <anlutsenko@xxxxxxxxx>
>> wrote:
>>
>>> Thank you Seth,
>>>
>>> Any chance this can be pushed to 5.0 branch too?
>>>
>>> > Keep in mind that the SWIG-based interface will (hopefully) be
>>> retired soon in favor of a fully-abstracted KiCad scripting language.
>>>
>>> Oh. When that happens there hopefully will be a transition period to
>>> give plugin writers a chance to migrate to this new language?
>>> When you say fully-abstracted Kicad language, that means it's own syntax
>>> and complete isolation from existing vast ecosystem of libraries that
>>> python has?
>>> Or is some interoperability planned?
>>>
>>> Regards,
>>> Andrew
>>>
>>> On Mon, Jul 30, 2018 at 8:53 AM Seth Hillbrand <seth@xxxxxxxxxxxxx>
>>> wrote:
>>>
>>>> This compiles cleanly and does not appear to have negative side effects
>>>> so I pushed it.  I made a small change to the boilerplate in math.i.
>>>>
>>>> Keep in mind that the SWIG-based interface will (hopefully) be retired
>>>> soon in favor of a fully-abstracted KiCad scripting language.  But since
>>>> that goal is still in the future, this patch may be useful in the interim.
>>>>
>>>> Thank you for your contribution to KiCad.
>>>> -S
>>>>
>>>> Am Di., 24. Juli 2018 um 20:30 Uhr schrieb Andrew Lutsenko <
>>>> anlutsenko@xxxxxxxxx>:
>>>>
>>>>> Hello,
>>>>>
>>>>> Can someone take a look at my patch please?
>>>>> It's my first so if I didn't submit it correctly please bear with me,
>>>>> I believe I did everything according to developers documentation.
>>>>>
>>>>> It would be great to see this in next release. Patch applies cleanly
>>>>> to both 5.0 branch and master.
>>>>>
>>>>> Regards,
>>>>> Andrew
>>>>>
>>>>> On Sun, Jul 22, 2018 at 6:05 PM Andrew Lutsenko <anlutsenko@xxxxxxxxx>
>>>>> wrote:
>>>>>
>>>>>> I figured it out, please see attached patch for the fix.
>>>>>>
>>>>>> On Sun, Jul 22, 2018 at 3:46 AM Andrew Lutsenko <anlutsenko@xxxxxxxxx>
>>>>>> wrote:
>>>>>>
>>>>>>> I tried adding this in kicad.i
>>>>>>>
>>>>>>> #include <geometry/shape_line_chain.h>
>>>>>>> %include <geometry/shape_line_chain.h>
>>>>>>>
>>>>>>> It compiles and SHAPE_LINE_CHAIN type shows up in pcbnew.py
>>>>>>>
>>>>>>> But I need VECTOR2I as well to get the actual coordinates. However
>>>>>>> adding math/vector2d.h header in a similar way leads to following
>>>>>>> linking
>>>>>>> error that I am not sure how to fix:
>>>>>>>
>>>>>>> D:/devtools/msys64/mingw64/bin/../lib/gcc/x86_64-w64-mingw32/7.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe:
>>>>>>> CMakeFiles/pcbnew_kiface.dir/objects.a(pcbnew_wrap.cxx.obj):pcbnew_wrap.cxx:(.rdata$.refptr._ZN7VECTOR2IiE10ECOORD_MINE[.refptr._ZN7VECTOR2IiE10ECOORD_MINE]+0x0):
>>>>>>> undefined reference to `VECTOR2<int>::ECOORD_MIN'
>>>>>>> D:/devtools/msys64/mingw64/bin/../lib/gcc/x86_64-w64-mingw32/7.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe:
>>>>>>> CMakeFiles/pcbnew_kiface.dir/objects.a(pcbnew_wrap.cxx.obj):pcbnew_wrap.cxx:(.rdata$.refptr._ZN7VECTOR2IiE10ECOORD_MAXE[.refptr._ZN7VECTOR2IiE10ECOORD_MAXE]+0x0):
>>>>>>> undefined reference to `VECTOR2<int>::ECOORD_MAX'
>>>>>>>
>>>>>>> Can someone who understands swig help me figure this out?
>>>>>>>
>>>>>>> On Sat, Jul 21, 2018 at 12:36 PM Andrew Lutsenko <
>>>>>>> anlutsenko@xxxxxxxxx> wrote:
>>>>>>>
>>>>>>>> Hello Kicad team,
>>>>>>>>
>>>>>>>> I'm writing a pcbnew action plugin that needs to read footprint pad
>>>>>>>> coordinates and shape.
>>>>>>>> I had no trouble figuring out all shapes except for custom one. It
>>>>>>>> is
>>>>>>>> represented by SHAPE_POLY_SET which has a corresponding python type
>>>>>>>> but to
>>>>>>>> get actual vertex coordinates you have to either call one of
>>>>>>>> Iterate()
>>>>>>>> variants or get polygon outlines with Outline(). They both return
>>>>>>>> objects
>>>>>>>> (ITERATOR and SHAPE_LINE_CHAIN) that are pointers and have no useful
>>>>>>>> attributes. As far as I understand they just don't have
>>>>>>>> corresponding swig
>>>>>>>> bindings so they are not exposed properly in python.
>>>>>>>>
>>>>>>>> Am I missing some obvious way to read custom pad polygon data in
>>>>>>>> python?
>>>>>>>>
>>>>>>>> If not, what would it take to add that?
>>>>>>>>
>>>>>>>> I don't understand much about how swig configuration works but
>>>>>>>> cursory
>>>>>>>> glance at Kicad source suggests that there needs to be an import for
>>>>>>>> shape_line_chain.h in common/swig/kicad.i, similar to
>>>>>>>> shape_poly_set import.
>>>>>>>>
>>>>>>>> If you are curious about what the plugin does, here are some of it's
>>>>>>>> results, generated from Kicad demos:
>>>>>>>> http://plugindemo.bitballoon.com/1/
>>>>>>>> http://plugindemo.bitballoon.com/2/
>>>>>>>> http://plugindemo.bitballoon.com/3/
>>>>>>>>
>>>>>>>> #2 above has a solder bridge with custom shape pads that are not
>>>>>>>> rendered.
>>>>>>>> Links are short lived.
>>>>>>>>
>>>>>>>> Regards,
>>>>>>>> Andrew
>>>>>>>>
>>>>>>> _______________________________________________
>>>>> 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
>>>>>
>>>> _______________________________________________
>> 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
>>
>