Re: Building GCC 8 using 32-bit addressing on a 64-bit machine
Gordon Steemson via Gcc-help <[email protected]> Tue, 28 Apr 2026 07:52:13 -0700
| Newsgroups | gmane.comp.gcc.help |
|---|---|
| Message-ID | <[email protected]> |
> On Apr 28, 2026, at 7:32=E2=80=AFAM, Jonathan Wakely <[email protected]= m> wrote: >=20 > =EF=BB=BFOn Tue, 28 Apr 2026 at 15:31, Jonathan Wakely <jwakely.gcc@gmail.= com> wrote: >>=20 >>> On Tue, 28 Apr 2026 at 09:59, Gordon Steemson via Gcc-help >>> <[email protected]> wrote: >>>=20 >>> You need a relatively recent GCC, like GCC 8 or 9, to compile current ve= rsions thereof. The system compiler on Mac OS 10.5 is Apple's build of GCC _= 4.2_. Building a version of GCC in the vicinity of 8 is a necessary interme= diate step. >>=20 >> You can build GCC using Clang, it doesn't have to be GCC. >=20 > Ah, but it looks like Mac OS 10.5 doesn't have Clang either! Wow, just > stuck with GCC 4.2 :-) Correct. Also, Clang has never built software for 64-bit PowerPC, because A= pple accidentally screwed up the implementation thereof (the error-unwinding= data that should be prepended to each function gets mangled into a single w= ord instead of being laid out as documented), and they took advantage of the= transition to Intel to declare that they refused to fix it. This is also w= hy the community-driven "Mac OS 10.6 'Snow Leopard' on PowerPC" effort is 32= -bit only.=