Comment 49 for bug 992124

I guess so. Will we have to open a bug for the lack of OpenMP support in
Inkscape, or is that considered a clang bug or a bug in OpenMP itself?

On Thu, Apr 11, 2013 at 3:38 PM, Alex Valavanis
<email address hidden>wrote:

> I wonder if we have a clear goal with this request. If it's a case of
> just being able to compile with clang, then I think the report can be
> closed now. Similarly, if it's a case of being able to compile
> "cleanly", I think we're about as clean as with GCC.
>
> As far as I can tell, the only remaining issue is that clang doesn't
> offer OpenMP support... and there's nothing we can do about that other
> than wait for upstream; Inkscape will compile and run, but with some
> performance loss.
>
> Can we mark this as "Fix Committed"?
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/992124
>
> Title:
> RFE: clang support
>
> Status in Inkscape: A Vector Drawing Tool:
> In Progress
>
> Bug description:
> Follow-up report to
> Bug #889172 “Compile failure (clang): "variable length array" errors in
> box3d.cpp”
> <https://bugs.launchpad.net/inkscape/+bug/889172>
>
> Rationale:
> - «clang support is a good thing»
> <
> http://thread.gmane.org/gmane.comp.graphics.inkscape.devel/36760/focus=36767
> >
>
> - OS X is migrating from GCC to CLANG (latest Xcode versions >= 4.2 no
> longer include Apple's GCC at all). It might be desirable to support
> building Inkscape using the system compiler on latest OS X versions
> (compiling Inkscape with custom GCC versions can cause known
> compatibility issues because the C++ libraries are built with a
> different compiler version [1]).
>
> Drawbacks:
> - Missing OpenMP support (Clang 3.x)
>
> Current status:
> Even though most of the reported 'non-POD element type' issues have been
> fixed in current trunk (and backported to 0.48.x), Inkscape still fails to
> build with clang:
>
> 1) autotools - configure.ac: configure fails with
> > checking GNU compiler version...
> > configure: error: gcc >= 3.0 is needed to compile inkscape
> The check for the compiler version has 'gcc version' hardcoded and fails
> with clang. [2]
>
> 2) unsolved 'variable length array of non-POD element type' errors
> - src/2geom/basic-intersection.cpp
> - src/2geom/solve-bezier-parametric.cpp
>
> 3) blocker: failure to compile 'src/color-profile.cpp'
> <http://www.graphicall.org/ftp/ideasman42/inkscape_clang.html>
>
> «I think the problem is that ColorProfileImpl should be defined in the
> Inkscape namespace instead of in the global namespace, so that it
> matches its declaration in color-profile.h. The code has an using
> declaration that pulls ColorProfileImpl into the global namespace and
> it looks like GCC then allows defining the type in the scope of the
> using declaration, but I doubt whether this is standards conforming.» [3]
>
> Test system:
> OS X 10.7.2, MacPorts 2.0.4 (up-to-date)
> Inkscape 0.48+devel r11297
>
> Clang versions tested:
> - Apple clang version 3.0 (tags/Apple/clang-211.12) (based on LLVM
> 3.0svn)
> - clang version 3.0 (tags/RELEASE_30/final) (MacPorts)
> - clang version 3.1 (branches/release_31 155781) (MacPorts)
>
> ---
> [1] <
> http://wiki.inkscape.org/wiki/index.php/FAQ#On_Linux.2C_Inkscape_crashes_with_.22invalid_pointer.22_message
> >
> [2] <http://trac.macports.org/ticket/31492>
> <
> http://fink.cvs.sourceforge.net/fink/dists/10.7/stable/main/finkinfo/graphics/inkscape.info?view=markup#l104
> >
> <http://www.freebsd.org/cgi/query-pr.cgi?pr=161204>
> <
> http://clang.debian.net/status.php?version=3.0&key=WRONG_GCC_ASSUMPTION>
> [3] <
> http://thread.gmane.org/gmane.comp.graphics.inkscape.devel/37024/focus=37060
> >
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/inkscape/+bug/992124/+subscriptions
>