Re: gEDA: Icarus Verilog snapshot 20030308

[Trevor Williams <trevorw-at-NOSPAMcharter_dot_net>]

Stephen Williams wrote:

>It's been a while since the last snapshot, the result being a
>*lot* of changes. The highlights are:
>
>    New interactive mode,
>    Compiler performance improvements,
>    Repaired Mingw port
>
>The new interactive mode in particular has the capacity to inspire
>a lot of interesting add-on projects.
>
>Anyhow, y head is swimming with big projects affecting simulation,
>including but not limited to:
>
>    Changing how parameters are handled (for the sake of SWIFT),
>    Trying to push some specify block support closer to vvp,
>    real valued nets, and
>    rewiring of vvp functor structures in support of the above.
>    Performance!
>
>I've been getting a lot of mail recently that has also been
>keeping me busy. I'm trying to keep up, but don't feel slighted
>if I occasionally ask you to repeat yourself after some delay
>getting back to you.
>
>Here are the snapshot files:
>
>   <ftp://icarus.com/pub/eda/verilog/snapshots/verilog-20030308.tar.gz>
>   <ftp://icarus.com/pub/eda/verilog/snapshots/verilog-20030308.txt>
>
>
>Release Notes for Icarus Verilog Snapshot 20030308
>
>The vvp command has gained an interactive mode. The $stop system task,
>or Ctrl-C, will enter this interactive mode, from which you can
>traverse the scopes of the design and invoke system tasks. You can
>also do a clean finish from interactive mode, useful when you want to
>stop the simulation with a clean LXT output. You can even resume the
>simulation. This is enough infrastructure to support interactive
>debuggers written in VPI.
>
>IEEE1364-2001 style PLI_* types are being introduces, and the PLI
>header files are gradually being adjusted to use these new types. Even
>though some of the type choices are pretty weird, these are better
>then the purely random choices from the -1995 standard.
>
>The latest lxt updates from Tony Bybell have been incorporated. This
>introduced new dependencies on bzlib2 and zlib, but when full
>compression is turned on, the output LXT files are very compact. Very
>competitive with commercial tools. This update also introduces
>dead-zone support to LXT, so the later gtkwave-1.* and gtkwave-2.*
>viewers will graphically show regions where dumps were temporarily
>suspended. See the vvp man pages for options.
>
>A new dangling port warning has been added to the compiler. This has
>been successfully used to catch incorrectly connected modules in large
>designs.
>
>Some code generator bugs relating to system functions as arguments to
>system tasks have been fixed. This should now work in general. VPI
>calltf functions will see constant values for the evaluated argument,
>but it will be the correct value.
>
>Some minimal progress towards specify block support has been
>made. Nothing particularly visible yet, except that some bugs in the
>BNF for specify paths have been fixed.
>
>The mingw port has been fixed, and the mingw.txt instructions greatly
>enhanced. Compiling under Windows has always been harder then
>compiling elsewhere, so detailed instructions can always be made
>better.
>
>Various bugs in real value support were fixed, including the addition
>of $realtobits. (Still no $bitstoreal.) IEEE754 support is in pretty
>good shape, actually. Still some glitches and gaps, but useable.
>
>A bunch of acc_ functions were added since the last snapshot. This
>improves compatibility with Cadence compatible PLI1 modules, still
>working towards functional SWIFT support. Also, acc_ string table
>support has been added. This better matches the behavior of commercial
>tools.
>
>Internally, the handling of names by the compiler has been drastically
>rewritten. This leads to some minor changes to the ivl_target api, and
>considerably less memory use during compile. Compilation is a bit
>faster now, as well. These changes also improve memory usage at run
>time, so expect some minor performance enhancements there.
>
>Minor makefile fixes to account for parallel builds.
>
>  
>
Stephen,

I downloaded this snapshot and attempted to compile, but got the 
following compiler error:

c++ -DHAVE_CVS_IDENT=1 -DHAVE_CONFIG_H -g -O2 -I. -I. -MD -c parse.cc -o 
parse.o
parse.cc: In function `int VLparse()':
parse.cc:6887: ISO C++ forbids declaration of `yylerrsp' with no type
parse.cc:6887: conflicting types for `int yylerrsp'
parse.cc:3105: previous declaration as `vlltype*yylerrsp'
parse.cc:6887: invalid conversion from `vlltype*' to `int'
parse.cc:6888: invalid type argument of `unary *'
parse.cc:6930: invalid operands of types `int' and `vlltype*' to binary `
   operator-'
parse.cc:6930: invalid operands of types `int' and `vlltype*' to binary `
   operator-'

I am using version "gcc version 3.2 20020903 (Red Hat Linux 8.0 3.2-7)" 
of the c++ compiler and version "bison (GNU Bison) 1.875" of bison.  Any 
ideas?

Thanks,
Trevor