February 28, 2006 Coping with a broken math lib ? | ||||
|---|---|---|---|---|
| ||||
... or at least one that lacks several "long double" functions, such as the one in Darwin 8.0.1 for x86: http://www.opensource.apple.com/darwinsource/10.4/Libm-92/ It's OK for "powerpc", but the "i386" version wasn't up to speed - so you got linker errors for missing symbols... http://www.algonet.se/~afb/d/gdc-darwin8-x86-errors.txt In other words, it's just like it says in std.c.math module: private import gcc.config; // unfortunately, these do not always exist in a library It's been "fixed" in a newer version of Libm (in Darwin 8.4+), but it would be "nice" if GDC could detect the missing ones ? Perhaps fall back to using a "double" version, or other such workaround ? The "real" type isn't 80 bits everywhere anyway. I also fixed a ppc conditionalizing in the ldcompat part of it: http://www.algonet.se/~afb/d/ldblcompat.d (a fix for i386/ppc64) If you hack the references to all the missing math functions out, *then* GDC works OK for Darwin 8 x86. Also works for Darwin 8 ppc. But since Darwin 8.0.1 "ships" with gcc-4061 (i.e. gcc-4.0.0 based), any 4.x based GDC needs to be built using: CC=gcc-3.3 CXX=g++-3.3 The same reason why gdcmac requires Xcode 2.2+, for Mac OS X 10.4... (There won't be any Darwin binaries in the gdcmac project, just Mac) --anders PS. Just wanted to know if it worked on "plain" Darwin too... It did. :-) | ||||
Copyright © 1999-2021 by the D Language Foundation
Permalink
Reply