Re: Bison generating C++

[Hans Aberg]

At 09.57 -0500 0-10-19, Davy Durham wrote:
[I didn't see your email cc'ed to bison-help, so I teke the liberty doing
that.]
>    Oh ok... well, I don't see any reason it needs to be in C++
>anyway..... I don't see how classes wouldn't really help in such an
>algorithmicly oriented application...  However, Some of the C++ features
>might be nice, like declaring vars anywhere you want (closer to the use of
>them I think is much better) and perhaps references could be useful...

Since Akim said he found it hard to concentrate on the C++ stuff for now,
perhaps we should take a rain-check on this discussion.

But in the first hand, people who already use C++, want an output
compatible with current C++. Extras, such as a nice syntax, classes,
namespaces, and polymorphy are add-on's from that point.

>    In just thinking about it writing C++ output:  Have you ever seen
>flex's C++ output.... I really didn't care for it... There was a C++ stub
>that declared the class or something, and flex's output included it (I
>don't remember exactly)...
>    I would have preferred it generate a stand-alone class not depentant
>on another other files (that was redundant)...   I think it would be nice
>for bison to create a .cc and a .h, and if you vi the .h file you can
>clearly see the public methods available to you at the beginning of the
>class definition, which itself is even near the top of the file..

I am using Flex and Bison together in that way. This is one reason Flex and
Bison needs some devlopment synchronization.

  Hans Aberg