Re: bison and C++

[Hans Aberg]

On 13 Dec 2006, at 01:11, Satya wrote:

> Putting the in the .y file would not suffice, as those functions need
> to be put in a file compiled by the C++ associated C-compiler;
>
> Well, when Bison generates an xyz.tab. c file, it can be compiled  
> by a C++ compiler. Thats the idea.

Once upon the time, Bison had an informal support for the C-parser to  
be compiled as C++, but this was dropped, being too difficult to  
maintain. Instead, a separate C++ skeleton was developed.

So, for C++, it is best to use the C++ skeleton file.

> one
> cannot then use C++ containers. If the C-parser is compiled using a C+
> + compiler, would it succeed, in view of that support for this has
> been terminated, then the parser stack will not invoke copy
> constructors when reallocating: it will only work with POD semantic
> types.
>
>
> The parser stack will not allow any special types anyway.

Sure it will: the C++ parser uses some 'std' container, like  
std::vector.

> You can put only *pointers* to STL or user defined data types on  
> the stack. I don't see how reallocation of the stack can affect  
> this. Maybe I am missing something. Can you elaborate pls.. (maybe  
> an example?)

It is the use of %union that forces POD semantic types, as it is  
implemented using a C++ 'union', which only admits those. Then one  
will have to tweak Bison to get it working.

  Hans Aberg