[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [Pkg-haskell-commits] darcs: haskell-hashable: Disable calls to mmap in testsuite, whose failure prevented build



Le 23. 10. 13 23:14, Joachim Breitner a écrit :
are you sure that that Cabal uses the same architecture names as Debian
does?

Greetings,
Joachim


Good morning,
That's a tough question. After investigation, I found that armel and armhf didn't exist and that arm was probably their synonym. Here are all the possible values for arch: alpha, arm, hppa1_1, hppa, i386, ia64, m68k, mipseb, mipsel, mips, powerpc64, powerpc, rs6000, s390x, s390, sparc64, sparc, vax, x86_64

I would still like to check the value of System.Info.arch on an arm* system.

Other remarks: x86_64 matches amd64 and Debian doesn't ship on hppa*, m68k, mipseb, powerpc64, rs6000, sparc64 and vax.

Reasoning :
Cabal takes its info from System.Info (http://www.haskell.org/cabal/users-guide/developing-packages.html#conditions). You can see a preprocessor token HOST_ARCH coming from ghcplatform.h in the source (http://hackage.haskell.org/package/base-4.6.0.1/docs/src/System-Info.html). Let's cd <ghc source tree>; includes/ghcplatform.h gets generated by includes/ghc.mk, line 84, looking for the arch in make variable TargetArch_CPP. Which is defined and passed on by configure, line 3681. I give up swimming up to the definition of target_cpu, partly because I shouldn't have been reading this generated script in the first place, but I can nevertheless deduce the set of possible values for arch.

Louis

--
When I grow up,
I will run a Tor node around the clock,
To help threatened protesters 'round the world.


Reply to: