Porting GMP v4.2.1 to QNX4 with Watcom C v10.6B

mts.spb.suxx at mail.ru mts.spb.suxx at mail.ru
Thu Aug 24 13:49:28 CEST 2006


  It was relatively easy to port GMP v4.1.4 but now I'm stuck with
  some strange errors while porting v4.2.1

  1) I did "./configure --disable-shared --host=none" and it completed
  2) I did "make" and it too completed OK, so it seems. All libraries
  are in place.
  3) When I do "make check" it fails with lots of unresolved symbols
  (see the attached "log.log1").

  If I have patience to compile the tests manually - providing the
  "cc" with the correct paths to all relevant libraryes - all goes
  well, all (compiled so far) tests pass and I can get to
  "mpn/hamdist" test which fails even with all the libs included. It
  reports an unresolved "___gmpn_popcount_" symbol.

  I look up the GMP v4.1.4 to see what is different against v4.2.1,
  the older library has the two different files -
  "mpn/generic/hamdist.c" and "mpn/generic/popcount.c", while the
  newer library introduces just one with "#ifdef"s to cover both

  If it was not for some weirdness in my tool-chain I'd try the
  splitted "popcount v4.2.1" and "hamdist v4.2.1" (see both attached,
  I need to know if I did [not] broke them). But my patience left me
  for now...

  I wonder what may be the cause of "libtool" producing a strange
  command line for "cc" with the "libgmp.a" library mentioned twice
  while all others are omitted?
  (By "all others" I understand these: libmpn.a, libmpz.a, libmpf.a,
  libmpq.a. Should they exist separately or should all their functions
  be present in the libgmp.a as well?)

  I'm sorry for too many questions in one letter, but please help me!

Best regards,
 Tony                  mailto:mts.spb.suxx at mail.ru
-------------- next part --------------
A non-text attachment was scrubbed...
Name: my_problems.zip
Type: application/octet-stream
Size: 3186 bytes
Desc: not available
Url : http://gmplib.org/list-archives/gmp-discuss/attachments/20060824/828ee9a7/attachment.obj 

More information about the gmp-discuss mailing list