Yes, if you find ports that need their compiler changed (or rather, need certain compilers blacklisted), please let us know. Overriding the compiler on the command line can be a useful troubleshooting step for identifying the solution to a bug, but should not be required in the general case; we should fix the portfile instead.
> On Jan 12, 2017, at 20:40, Richard L. Hamilton <[email protected]> wrote: > > Forgot I had +universal as the default - tried again with port install > clang-3.8 -universal, to get the pre-built versions, much quicker (and less > surprises from the road less traveled). I don't mind for libraries, but for > a compiler and long toolchain, that's nuts, esp. if there's no advantage. > > After that (minutes rather than hours!), gnutls built successfully, and > rev-upgrade found no linking errors. I've added that to my notes of what > might require some manual intervention, which currently require (given the > compilers I have installed) the following: > kerberos5 gcc-4.2 > dpkg clang > rrdtool clang > libgcrypt llvm-gcc-4.2 > gnutls macports-clang-3.8 > > I have a script that does a port upgrade minus certain blacklisted ports > (those and one or two others), which I deal with by hand, so that they don't > cause upgrading the rest to fail. Of course, on a newer OS, this just isn't > a problem. :-) But that old 2007 Mac Mini can't go past Lion anyway, and > Snow Leopard has the advantage of being the last version that can still do > Rosetta, so there's no incentive for me to upgrade it to Lion and lose that > capability. > > Perhaps some of those will now no longer be needed, if the corresponding > ports find and use the macports-clang-3.8 by default, if available. I'll try > to remember to try them in the future without specifying a compiler, to see > what happens. > > Thanks for the assist! > >> On Jan 12, 2017, at 20:01, Richard L. Hamilton <[email protected]> wrote: >> >> It'll likely be a few hours before (if!) the dependency chain for clang-3.8 >> finishes building, assuming I have sufficient disk space. :-) But what the >> heck, the machine is doing most of the work, so I'll try. >> >>> On Jan 12, 2017, at 19:54, Marius Schamschula <[email protected]> wrote: >>> >>> Richard, >>> >>> Please look at the discussion on trac: >>> >>> https://trac.macports.org/ticket/53295 >>> >>>> On Jan 12, 2017, at 6:49 PM, Richard L. Hamilton <[email protected]> wrote: >>>> >>>> From what I can find >>>> https://en.wikipedia.org/wiki/Advanced_Vector_Extensions >>>> the Core 2 Duo does _not_ support AVX (it's way too old - this machine was >>>> 2007). CPU info (from sysctl): >>>> hw.optional.floatingpoint: 1 >>>> hw.optional.mmx: 1 >>>> hw.optional.sse: 1 >>>> hw.optional.sse2: 1 >>>> hw.optional.sse3: 1 >>>> hw.optional.supplementalsse3: 1 >>>> hw.optional.sse4_1: 0 >>>> hw.optional.sse4_2: 0 >>>> hw.optional.x86_64: 1 >>>> hw.optional.aes: 0 >>>> machdep.cpu.max_basic: 10 >>>> machdep.cpu.max_ext: 2147483656 >>>> machdep.cpu.vendor: GenuineIntel >>>> machdep.cpu.brand_string: Intel(R) Core(TM)2 CPU T7200 @ 2.00GHz >>>> machdep.cpu.family: 6 >>>> machdep.cpu.model: 15 >>>> machdep.cpu.extmodel: 0 >>>> machdep.cpu.extfamily: 0 >>>> machdep.cpu.stepping: 6 >>>> machdep.cpu.feature_bits: 3219913727 58301 >>>> machdep.cpu.extfeature_bits: 537921536 1 >>>> machdep.cpu.signature: 1782 >>>> machdep.cpu.brand: 0 >>>> machdep.cpu.features: FPU VME DE PSE TSC MSR PAE MCE CX8 APIC SEP MTRR PGE >>>> MCA CMOV PAT PSE36 CLFSH DS ACPI MMX FXSR SSE SSE2 SS HTT TM PBE SSE3 >>>> DTES64 MON DSCPL VMX EST TM2 SSSE3 CX16 TPR PDCM >>>> machdep.cpu.extfeatures: SYSCALL XD EM64T LAHF >>>> machdep.cpu.logical_per_package: 2 >>>> machdep.cpu.cores_per_package: 2 >>>> machdep.cpu.microcode_version: 199 >>>> machdep.cpu.mwait.linesize_min: 64 >>>> machdep.cpu.mwait.linesize_max: 64 >>>> machdep.cpu.mwait.extensions: 3 >>>> machdep.cpu.mwait.sub_Cstates: 139808 >>>> machdep.cpu.thermal.sensor: 1 >>>> machdep.cpu.thermal.dynamic_acceleration: 0 >>>> machdep.cpu.thermal.invariant_APIC_timer: 0 >>>> machdep.cpu.thermal.thresholds: 2 >>>> machdep.cpu.thermal.ACNT_MCNT: 1 >>>> machdep.cpu.thermal.core_power_limits: 0 >>>> machdep.cpu.thermal.fine_grain_clock_mod: 0 >>>> machdep.cpu.thermal.package_thermal_intr: 0 >>>> machdep.cpu.thermal.hardware_feedback: 0 >>>> machdep.cpu.thermal.energy_policy: 0 >>>> machdep.cpu.arch_perf.version: 2 >>>> machdep.cpu.arch_perf.number: 2 >>>> machdep.cpu.arch_perf.width: 40 >>>> machdep.cpu.arch_perf.events_number: 7 >>>> machdep.cpu.arch_perf.events: 0 >>>> machdep.cpu.arch_perf.fixed_number: 0 >>>> machdep.cpu.arch_perf.fixed_width: 0 >>>> machdep.cpu.cache.linesize: 64 >>>> machdep.cpu.cache.L2_associativity: 8 >>>> machdep.cpu.cache.size: 4096 >>>> machdep.cpu.tlb.inst.small: 128 >>>> machdep.cpu.tlb.inst.large: 8 >>>> machdep.cpu.tlb.data.small: 16 >>>> machdep.cpu.tlb.data.small_level1: 256 >>>> machdep.cpu.tlb.data.large: 16 >>>> machdep.cpu.tlb.data.large_level1: 32 >>>> machdep.cpu.address_bits.physical: 36 >>>> machdep.cpu.address_bits.virtual: 48 >>>> machdep.cpu.core_count: 2 >>>> machdep.cpu.thread_count: 2 >>>> >>>> It doesn't even have SSE4, so it wouldn't make sense that it had AVX. >>>> >>>> So...how do I get it to run configure with the >>>> --disable-hardware-acceleration switch; do I have to edit the portfile, or >>>> is there something I can do from the command line? The following does NOT >>>> work (looking at the log file shows that the added arg was NOT passed): >>>> port upgrade gnutls configure.args-append=--disable-hardware-acceleration >>>> >>>> Now maybe I could dig the full list of normal args out of the portfile, >>>> and pass configure.args="[whole list of args]" on the command line, but >>>> that's just as bad as editing the portfile, because I'd have to look at it >>>> each time. >>>> >>>> I'm guessing that the AVX support in gnutls came in with the big version >>>> jump from the previous 3.4.17 to 3.5.8. I think that for those with older >>>> hardware, it would be much friendlier if the portfile supported a variant >>>> to disable hardware acceleration. >>>> >>>> >>>>> On Jan 12, 2017, at 06:56, Marius Schamschula <[email protected]> >>>>> wrote: >>>>> >>>>> Richard, >>>>> >>>>> I checked on my Leopard box, and found that gnutls 3.5.8 properly built >>>>> there. So, it seems that we have an issue unique to Snow Leopard. >>>>> >>>>> There are two options: >>>>> >>>>> 1) try to find a compiler other than Apple’s gcc that will properly deal >>>>> with the AVX extensions >>>>> >>>>> or >>>>> >>>>> 2) use the —disable-hardware-acceleration switch >>>>> >>>>>> On Jan 10, 2017, at 8:35 PM, Richard L. Hamilton <[email protected]> >>>>>> wrote: >>>>>> >>>>>> sh-3.2# uname -a >>>>>> Darwin myeye.pri 10.8.0 Darwin Kernel Version 10.8.0: Tue Jun 7 >>>>>> 16:33:36 PDT 2011; root:xnu-1504.15.3~1/RELEASE_I386 i386 >>>>>> sh-3.2# /usr/bin/gcc --version >>>>>> i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3) >>>>>> Copyright (C) 2007 Free Software Foundation, Inc. >>>>>> This is free software; see the source for copying conditions. There is >>>>>> NO >>>>>> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR >>>>>> PURPOSE. >>>>>> >>>>>> >>>>>>> On Jan 10, 2017, at 19:53, Marius Schamschula <[email protected]> >>>>>>> wrote: >>>>>>> >>>>>>> Richard, >>>>>>> >>>>>>> Unfortunately, I have no way of testing under Snow Leopard, even though >>>>>>> I test some builds under Leopard on a G5. >>>>>>> >>>>>>> It is clear the build is not finding the AVX extensions on you mini. >>>>>>> Given you have a Core2 Duo, the hardware is there. >>>>>>> >>>>>>> Can you give me the result of >>>>>>> >>>>>>> /usr/bin/gcc —version >>>>>>> >>>>>>> >>>>>>>> On Jan 10, 2017, at 6:17 PM, Richard L. Hamilton <[email protected]> >>>>>>>> wrote: >>>>>>>> >>>>>>>> Hardware Version: Mac mini (Mid 2007) >>>>>>>> Model Identifier: Macmini2,1 >>>>>>>> Mac OS X 10.6.8 10K549 (Snow Leopard) >>>>>>>> Xcode 3.2.6 DevToolsSupport-1806.0 >>>>>>>> MacPorts Version: 2.3.5 >>>>>>>> >>>>>>>> <main.log-gnutls.txt.gz> >>>>>>> >>>>>>> Marius >>>>>>> -- >>>>>>> Marius Schamschula >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>> >>>>> Marius >>>>> -- >>>>> Marius Schamschula >>>>> >>>>> >>>>> >>>>> >>>> >>> >>> Marius >>> -- >>> Marius Schamschula >>> >>> >>> >>> >> >
