Thank you, I had missed that message.
I tried this, and it worked! The entire process of building Poly/ML succeeded, including installation.
However, I am seeing strange crashes whenever I build HOL4, both in the 32- and the 64-bit versions.
Here is the end of the trace from the 32-bit build (not that you should understand HOL4 stuff), where it looks like compiling a file generates a bus error:
. . . Analysing boolScript.sml Compiling boolScript.sml Linking boolScript.uo to produce theory-builder executable Poly/ML 5.3 Enhanced Reporting Testing
Warning- in 'Holmake/ReadHMF.sml', line 108.
Matches are not exhaustive. Found near case first_special s' of NONE => error b "Unrecognised" | SOME = => (advance b, ... ...) | SOME : => ... ... (... ^ ...) Warning- in '/Users/palantir/hol/kananaskis-6/src/experimental-kernel/RawParse.sml', line 42. Matches are not exhaustive. Found near case cur of (NONE, _) => NONE | (SOME bslash, cs') => let in case ... ... of ... => ... | ... end | (... ..., ...) => ... ... (... ...) | (...) => ... ... Warning- in '/Users/palantir/hol/kananaskis-6/src/experimental-kernel/RawParse.sml', line 59. Matches are not exhaustive. Found near case cur of (NONE, _) => reduce_tm stk cur | (SOME (... ...), cs') => ... ... (... ...) (adv cs') | (... ..., ...) => ... ... (... ...) | (...) => ... ... Warning- in '/Users/palantir/hol/kananaskis-6/src/experimental-kernel/RawParse.sml', line 65. Matches are not exhaustive. Found near case stk of C t :: BK :: rest => reduce_tmb rest t cur | C t :: Start :: rest => SOME (t, ...) | ... ... :: ... :: ... => ... ... cur Warning- in '/Users/palantir/hol/kananaskis-6/src/experimental-kernel/HolKernel.sml', line 130. Pattern is not exhaustive. Found near val (h :: t) = List.rev alist Warning- in '/tmp/MLTEMPNU8qJX', line 1274. Pattern is not exhaustive. Found near val PM_LG {pfx = pfx, ifx = ...} = order <<HOL message: Created theory "bool">> /Users/palantir/hol/kananaskis-6/src/bool/boolScript: line 13: 63318 Bus error /usr/local/bin/poly <<'__end-of-file__' val _ = PolyML.Compiler.prompt1:=""; val _ = PolyML.Compiler.prompt2:=""; val _ = PolyML.print_depth 0; local val dir = OS.FileSys.getDir(); val _ = OS.FileSys.chDir (OS.Path.concat ("/Users/palantir/hol/kananaskis-6", "tools-poly")); val _ = use "poly/poly-init2.ML"; val _ = OS.FileSys.chDir dir; in end; val _ = List.map load ["boolScript"] handle x => ((case x of Fail s => print (s^"\n") | _ => ()); OS.Process.exit OS.Process.failure); __end-of-file__
Build failed in directory /Users/palantir/hol/kananaskis-6/src/bool
Just before the final error the computer seemed to heavily load the CPU for about 20-30 seconds, and then the error report was printed.
Michael, can you interpret these results?
Alternatively, is there some body of tests that I could run to validate that my Poly/ML installation is operating correctly?
Peter
On Wed, Sep 23, 2009 at 3:09 AM, David Matthews David.Matthews@prolingua.co.uk wrote:
Peter Vincent Homeier wrote:
With a completely fresh copy of the developer version, obtained by svn, I try to build Poly/ML in a 32-bit version under the latest version of Snow Leopard (OS X 10.6.1) by
cd polyml/polyml ./configure CFLAGS='-arch i386 -O3' CXXFLAGS='-arch i386 -O3' CCASFLAGS='-arch i386' make
Hasn't this been discussed already? ?See below from http://lists.inf.ed.ac.uk/mailman/private/polyml/2009-September/000585.html
David
David,
Yes, it works with --build=i386-apple-darwin.
Regards,
Rob.
On Wednesday 09 Sep 2009 5:55 pm, you wrote:
Rob, Try with ./configure --build=i386-apple-darwin CFLAGS='-arch i386 -O3' \ CXXFLAGS='-arch i386 -O3' CCASFLAGS='-arch i386'
With the change to config.guess the configure script is now picking up the default architecture as x86_64 rather than i386 which was what it was reporting before. ?Previously you had to override it in 64-bit mode; now you have to override it in 32-bit mode.
It would be good to have just a single option to select whether to build 32-bit or 64-bit on machines that support that. ?I can't see a good way to do it at the moment.
Regards, David