Re: member with constructor not allowed in union

[Hans Aberg]

At 14:18 +0100 2002/03/18, Akim Demaille wrote:
>Hans> I think that if you now have a C++ skeleton file, that should be
>Hans> put into the ordinary distribution. Those that want a C++ output
>Hans> should use that one, not compile the C output as C++.
>
>No.

Yes, this is what I think. :-)

>  Use CVS.

I think I will wait until it is sufficiently stable for putting in the
distribution. -- I think this is what others will do, so before that
happens, you will not get proper feedback.

>Hans> As for the C output only feature, I think one may need two
>Hans> skeleton files: One with a purely stack produces an overflow,
>Hans> and one purely dynamic. This is more memory efficient. But that
>Hans> could be made later at some point (with the C++ skeleton file in
>Hans> hand, there is no need for rush).
>
>I don't think anybody cares.

At least not I. :-)

>  If one does, she simply has to adjust
>the YYSTACK parameters, period.

If you would have followed your own RTFC dogma, you would have known that
this does not work. :-)

>  I have no intention to augment the
>maintenance burden.

As I said before, I will curiously await your solution for multi-language
output support. :-)

  Hans Aberg