kicad-developers team mailing list archive

[Brian Sidebotham <brian.sidebotham@xxxxxxxxx>]

On 31 October 2012 06:16, Miguel Angel Ajo Pelayo <miguelangel@xxxxxxx>wrote:

> Hi Brian, I'm not sure how dependent is wxPython on Cairo, if it's
> recommended as '0' but they enable it,
> may by it's needed. But it's just a feeling, 0-knowledge.
>
> Do you have some branch for this that I could replicate and play around to
> figure out what's happening?
> or it's tested manually at the moment?
>
> Also a binary package (with some debug info) could do the trick for me
>
> I'd like to trace that wxpython init call down, and find where does it
> stop (and why).
>
> Thanks a lot for your effort,
>
> Mike,
>
>
Hi Mike,

I had to stop looking at the build for a while, but I should get another
chance to look at it earlier next week. It would seem that we need to
compile Python with MinGW, and wxPython.

I had a quick look at the source for Inkscape and they have a separate
repository which includes a pre-built python for Windows which gets
included with their installer. They do not have it as a build process for
building Inkscape though, only as a separate pre-built binary:
https://launchpad.net/inkscape-devlibs

I suspect this may have to be the route that KiCad goes down in order to
package the new scripting functionality. At least it means that we can
totally control the python version on Windows.

I don't have a branch, but maybe I should create one. It is likely there
will have to be changes in order to build scripting support for Windows.
Even the find_package( stuff uses the registry (yuck!) entries to discover
where python is located and this will always be wrong because we are using
MinGW to build the KiCad binaries, not MSVC which the standard python
install is built with.

Best Regards, Brian.