Re: Bison development

[Akim Demaille]

>>>>> "Hans" == Hans Aberg <address@hidden> writes:

Hans> I recall that it was the fact that your C++ compiler finally
Hans> detected your illegal use of unions that caused your code to
Hans> break. -- We did not do any changes with respect to those
Hans> union's when altering the skeleton file. And even in the days
Hans> the code compiled, it was not safe.

I also have to use a modern C++ environment, which before was
optional.  Personally I don't care, I'm not developing a portable
application in this case.  People who do are in trouble now.

Hans> So I think it is you, like the Spartans (according to
Hans> Herodotus), that is trying to kill the messenger because you
Hans> don't like the message. :-)

Hm...  I understand what you mean here, and it might be the case
indeed.


>> I'll only remember that you don't offer your help where we need
>> some.

Hans> I can't recall any request for assistance.

There is the TODO list for instance.

Hans> No specs on the stuff you are doing has been announced on the
Hans> Bison lists what I recall, so it is hard for anyone to judge
Hans> what to assist with.

Fair enough.

Hans> Unless the intent is to run the Bison development as a business
Hans> project, in which case it would be more suitable paying people
Hans> accordingly. :-)

That remains a Free Software project.  I'd like to be paid to work on
it, but it isn't the case.



OK, if you want a project, then there is one thing we need: that
someone find a good implementation of bit vector to implement sets.
We need a good implementation, efficient and with a nice interface.
There is a crucial need in Bison, but for the time being, unions,
intersections etc. are computed but loops all over the place.  We need
macros/routines that raise the abstraction level up to sets.