Re: Reference variables in jawin?

Hi Roger,
Try ANTLR from http://www.antlr.org/
This tool is beeing developed for a number of years. I use it for a long
time. The code generated very easy to debug. You can build with it AST and
store it in XML format.

Vitaly Shelest



"Roger I Martin PhD" <[email protected]> wrote in message news:<[email protected]>...
> Hi Morten,
> 
> Do you think that generating for dll's from header files would be a good
> next implementation that I could do next?  I was looking at it this weekend
> because I had started on it before.  I was trying to see what was the best
> way to go about it.  Tried Chaperon and couldn't get it to work.  Looked at
> JFlex but don't see how to generate xml with it.  Tempted to simply write
> code that parses a header file and outputs xml description similar to what
> was done for type libs.
> 
> What do you or anybody think?
> 
> --Roger
> ----- Original Message -----
> From: "Morten Andersen" <[email protected]>
> To: <[email protected]>
> Sent: Monday, June 21, 2004 9:27 AM
> Subject: Re: [JAWIN] Reference variables in jawin?
> 
> 
> > Hi Jonas
> >
> > Jonas Lind?n wrote:
> > >>>
> > >>>This is how the C++ code looks like:
> > >>>int cbAIn( int BoardNum, int Channel, int Range, unsigned short
> > >>>*DataValue );
> > >>>
> > >>
> > >>Is the "DataValue" parameter used for returning a short value? Or is it
> > >>used for inputting a short array? So is it used as an [in] parameter or
> > >>an [out] parameter.
> > >>
> > >
> > >
> > > The DataValue is used for returning a variable of type short.
> >
> > Then I think something along the lines with this should do it:
> >
> >    FuncPtr fp = new FuncPtr("cbw32.DLL", "cbAInScan");
> >    NakedByteStream nbs = new NakedByteStream();
> >    LittleEndianOutputStream leos = new LittleEndianOutputStream(nbs);
> >    leos.writeInt(boardNum);
> >    leos.writeInt(channel);
> >    leos.writeInt(range);
> >
> >    // call the function with three ints and 4 bytes allocated for
> >    // the datavalue - see explanation of the instructionstring below
> >    byte[] result = fp.invoke("IIIA:I:L4L4L4n4", 16, nbs, null,
> >      ReturnFlags.CHECK_NONE);
> >
> >    LittleEndianInputStream leis = new LittleEndianInputStream(
> >      new ByteArrayInputStream(result));
> >
> >    // return val
> >    int retVal = leis.readInt();
> >
> >    // and datavalue - should perhaps not be casted to a short? - since
> >    // unsigned short in native code?
> >    short dataValue = (short)leis.readInt();
> >
> >
> > The instruction string "IIIA:I:L4L4L4n4" should be interpreted like this
> > (on the form XXX:Y:ZZZ , where XXX is directions for [in], Y is the
> > [retval] and ZZZ is [out]-directions) :
> > - I: in integer (or other 4 byte value).
> > - A: skip in instream and allocate 4 bytes on stack (yes, I know a short
> > is only 2 bytes, but on the stack the calling convention __stdcall works
> > in chunks of 4 bytes, so smaller parameters takes up 4 bytes).
> > - L4: on the output = skip 4 bytes (we do this for all the
> [in]-parameters).
> > - n4: read 4 bytes of the stack and return.
> >
> > I know this code is not for the faint at heart, and over time the Jawin
> > Type Browser should hopefully be able to generate it for DLL's too.
> >
> > I am working on documenting the instruction strings too (pt. there is
> > only this outdated documentation for this:
> > http://jawinproject.sourceforge.net/instruction_docs.html ).
> >
> > Please note that I tested this code on the CVS HEAD version of Jawin,
> > and there have been significant cleaning of the API for calling
> > DLL-entry points since the last released version. So maybe this code
> > does not work in the released version, if that is the case I urge you to
> > check out the CVS version and build it yourself (please see:
> > http://jawinproject.sourceforge.net/jawindeveloper.html ).
> >
> > In the CVS-HEAD I also started out on some detailed information about
> > calling DLL-entry points (see docs/jawinuserguide_dll.html in the
> > checked out source), although it is not complete on some of the finer
> > issues (as this is).
> >
> > Hope this helps you, and feel welcome to email again if you still have
> > issues with this.
> >
> > Best Regards
> > Morten
>