Compiling under Solaris 10

[Marius Konitzer]

I tried to compile inkscape (svn head) under Solaris 10 using gcc-4.2. I had to patch a few lines only (see attachment) and what can I say -- it worked like a charm!

Regarding is{nan,inf,finite,normal} we seem to be in the same situation as under Solaris 8 here and (assuming continuity) probably under Solaris 9, too.

What I don't understand is, why some stuff is fixed for gcc-3.2 only; I had to remove this condition to make it work for me. This issue merely looks OS-specific, so one would expect it to be quite independent from the used gcc version, or am I wrong?

[Joshua Blocher (verbalshadow)]

This Patch is svn as of revision 19126.

[Sam Sirlin (samuel-w-sirlin)]

From the comments it looks like it was fixed, but it
doesn't seem fixed as of 0.47...

[Marius Konitzer (m-konitzer)]

It was until someone awkwardly broke it again in rev 19127. So please re-patch src/2geom/isnan.h.

Looks like we needlessly have to keep on patching till 0.48 is released. *sigh*

[su_v (suv-lp)]

re-opening - can someone confirm the patch in comment #1 for 'src/2geom/isnan.h' still works, without causing issues in 2geom or on other platforms?

[Bryce Harrington (bryce)]

The patch from comment #1 no longer works, but really only the first chunk is required; the remainder of that patch is already applied on trunk.

The regression is probably because the original fix was applied to inkscape's codebase and not to upstream 2geom. I've re-applied the fix to trunk, but am making the same mistake of not upstreaming it...