Re: What are the benefits to migrate from Bison++ to Bison

[Timothy Madden]

On 06/29/2012 10:09 AM, Akim Demaille wrote:
> Le 28 juin 2012 à 17:28, Timothy Madden a écrit :
>
> > First, it looks like you want to migrate from old bison to the newer bison++.
>
> Why do you qualify Bison as old, and Bison++ as newer?  The
> latest release of Bison for instance is about one month old.
>
> > If that is so, it means you want to generate a bison C++ parser, instead of a 
> > bison C parser. This means your question is about whether or not to migrate 
> > from C to C++. It is not necessarily a topic for a bison discussion group, but 
> > it might still be considered somewhat related.
>
> Bison does produce C++ parsers.
>
> > Unlike in flex, the bison C++ interface is not marked as experimental, so you 
> > should be safe with it, but you should know that GLR parsers currently need the 
> > C interface from the old bison.
>
> That is true and false.  First, there is a GLR C++ parser,
> that works.  But it does not support the variants, that's
> the only issue.

Oh .. sorry I was under the impression they are actually the same 
program, with bison++ being a symbolic link to bison (on my CentOS 6.2 
flex++ is such a symlink to flex, so I figured it must be the same with 
bison, though I have no bison++ installed).

As I thought them to be the same, I only talked about bison being "old" 
and bison++ being "new" to keep a simplified view of the things for the 
OP, which looks to me like he wants a simplified view. By bison++ I 
meant "bison with the C++ interface", which I would say is "new" based 
on (in relation to) the fact that it emerged after the C interface.

About the GLR it just did not work for me, which I suspect is because 
distributions do not update their package as fast as one month, so there 
are still people who don't have the latest version, although they are 
up-to-date with their OS distribution.

Sorry for the confusion,
Timothy Madden