[Gm2] latest cvs changes

[Gaius Mulley]

Hi,

here are the latest cvs changes:


   * gm2/gm2-libs-pim/RealConversions.mod:  intemediate
     checkin of catching nans and other non fpnormals.
   * The following changes fix a bug reported by
     Scott Iverson
     (http://lists.gnu.org/archive/html/gm2/2010-01/msg00022.html)
   * gm2/gm2-compiler/M2GCCDeclare.mod:  improved
     debugging for variant tags.  Only walk variant tags
     which are necessary in the final production of the
     equivalent C struct passed to the GCC backend.
     Ignore dummy ELSE END inside variant records during
     the construction of the equivalent C struct.
   * gm2/gm2-compiler/SymbolTable.def:  new procedures
     (IsRecordFieldAVariantTag) and (IsEmptyFieldVariant).
   * gm2/gm2-compiler/SymbolTable.mod:  implement
     new procedures (IsRecordFieldAVariantTag) and
     (IsEmptyFieldVariant).

   * gm2/gm2builtins.c:  introduced __builtin_ilogbf,
     __builtin_ilogbl and __builtin_ilogbl.
   * gm2/gm2-libs/Builtins.def:  introduced definitions
     for ilogb, ilobl and ilobf.
   * gm2/gm2-libs/Builtins.mod:  implemented references
     for builtins.
   * gm2/gm2-libs/cbuiltin.def:  introduced non builtin
     definitions for ilogb, ilobl and ilobf.
   * gm2/gm2-libs-pim/RealConversions.mod:  doPowerOfTen
     reimplemented and added comment.  LongRealToString
     broken up into two procedures and it now tests for
     NaN.
   * gm2/gm2-libs-pim/RealConversions.def:  updated
     comment for LongRealToString.
   * gm2/gm2-libs-pim/RealConversions.mod:  doPowerOfTen
     reimplemented.

   * gm2/gm2-libs-pim/RealConversions.mod:  fixed bug in
     exponent length and ensure that no white space is
     created.  Corrected name of procedure
     SetNoOfExponentDigits.

regards,
Gaius