barracuda is correct about libstdc++ being the default on 10.6 for both x86_64 and ppc.
ryan even said this in a bugticket that our beloved self-anointed sage responded to: https://trac.macports.org/ticket/56042 `` 10.7 and later ship with libc++, but 10.6 doesn’t ‘' to criticise others as you did below (`` It is very very different from 10.6 / Intel / libc++. It builds with gcc, not clang. It links against libstdc++, nott libc++ ‘') without being aware of this, in spite of participating in a ticket that made it a focal point, is kind of funny. sort of enforces my previous comment about “not being suitable for the role you [have] covet[ed, probably for more than a decade]” good on you for standing up for yourself, barracuda. enough is enough. rootie can’t let someone bully others and ruin the enthusiasm of important contributors. speak up, rootie. > On Jan 24, 2025, at 10:14 PM, Sergey Fedorov <[email protected]> wrote: > > Ken, you got some degree of superficial acquaintance with the system in > question 2,5 years ago and keep convincing yourself and others that your, > obviously emotionally biased, take represents an accurate account of reality, > and that reality could not possibly have changed ever since. > > As a matter of fact, it is 10.6 on Intel that is completely hacked in > MacPorts, and uses C++ runtime which Apple never had on 10.6. There is no > such a thing as libc++ on 10.6. If you insist on ritualistic preservation of > whatever Apple did at the date of release of the OS, perhaps start by > removing your hacks from MacPorts which force an alien runtime and clang-11. > > Years pass, and you still cannot demonstrate all those hundreds of > 10a190-specific hacks in MacPorts tree, because they do not exist. They do > not exist in my fork either in that numbers or anything remotely close to > that. > > Yes, libstdc++ is used on powerpc, which, ironically, makes 10.6.8 on powerpc > closer to the original than on x86. Though libc++ works fine, once someone’s > (I won’t bother even checking whose) spaghetty hacks for clangs are fixed or > dropped. > > > > > > On Sat, Jan 25, 2025 at 11:53 AM Ken Cunningham > <[email protected]> wrote: > Well -- of course 10.6-PPC needs lots and lots and lots and lots of special > workarounds. > > 10.6-PPC is basically 10.5 PPC wearing lipstick and a wig. > > It is very very different from 10.6 / Intel / libc++. It builds with gcc, > not clang. It links against libstdc++, nott libc++. It does not have the 10.6 > kernel features or framework / library supports. It is much more like an > early version of 10.5, which is why you need special workarounds all over the > damn place for it to make it behave like 10.5 even though it reports itself > as 10.6 > > Which is why -- so far -- there are hundreds of workarounds for 10.6-PPC in > the Portfiles that go like this: > > if (10.6) but ! build.arch==ppc) { > do some normal thing > } > > Just complete and total garbage. > > And God only knows what crap commits have been forced on unsuspecting > upstreams for this nonsense. I shudder to think. > > K > Thanks, Gagan
