[PySide] Shiboken Large Address Bug

[Tony Barbieri]

Thanks for your response Luc-Eric!

I think that's what we'll do for the time being in our build, but it would
be great if it could make it into one of the upcoming releases officially.

Best,

-tony


On Thu, Feb 6, 2014 at 9:59 PM, Luc-Eric Rousseau <lucericml at gmail.com>wrote:

> we have used unsigned long long in our build (Autodesk Maya), as stated
> here:
> https://bugreports.qt-project.org/browse/PYSIDE-191
>
> you could make that signature ifdef'ed to only occur on windows x64 if
> you're worried about binary compatibility. There are very few 64-bit
> python/qt/pyside users on Windows, as proven by how long this code has
> been around.   Nobody else is affected.  But on windows x64, it's
> totally broken and the bug will happen all the time on WIndows 8,
> because it's started to allocate memory above the 4 gig fold by
> default.
>
> On Thu, Feb 6, 2014 at 6:52 PM, Tony Barbieri <greatrgb at gmail.com> wrote:
> > Not the most elegant solution but it seems that unsigned long long is
> > already an acceptable primitive supported by shiboken.  I really have no
> > clue how to add in new primitive types or the implications that involves.
> > Would this be a suitable solution for the next official release until
> more
> > time can be devoted to fixing it properly?
> >
> >
> > On Thu, Feb 6, 2014 at 4:37 PM, Matthew Woehlke
> > <mw_triad at users.sourceforge.net> wrote:
> >>
> >> On 2014-02-06 16:04, Tony Barbieri wrote:
> >> > It would be great if we could get this fixed.  It's kind of a biggie
> for
> >> > us.  I would help out but I really don't know where to start.  If I
> find
> >> > the time I'll try to take a crack at it.
> >>
> >> Alas, the same (both your knowledge and time points) is true for me :-).
> >> At least, I might have a better /general/ knowledge of the code base,
> >> but not these specifics.
> >>
> >> That said, my main concern would be what (if any) compatibility
> >> implications a change will have (e.g. I think it may require that
> >> existing bindings are regenerated and rebuild, or else they will have
> >> runtime errors), and to what extent that's a concern vs. acceptable to
> >> address the issue. (Maybe leaving the old function, marked deprecated,
> >> and adding an overload with the correct signature could fix the issue
> >> without a BIC.)
> _______________________________________________
> PySide mailing list
> PySide at qt-project.org
> http://lists.qt-project.org/mailman/listinfo/pyside
>



-- 
-tony
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/pyside/attachments/20140207/2594cb3f/attachment.html>