On Tue, Dec 02, 2003 at 09:17:31AM +0000, Angus Leeming wrote:
> Andre Poenitz wrote:
> > At least making the LyXLex interface a bit similar to the
> > std::streams would make me feel better.
> >
> > Patch attached.
>
> > + /// file is open and end of file is not reached
> > + operator void *() const;
>
> I don't like this.
But that's what std::iostream does.
> We already have problems catching implicit
> conversions. Witness all those (bv()) arguments that are slowly being
> changed to (true) or (false) and have nothing to do with the
> existence of a BufferView anymore.
>
> > From a less political, more coding, point of view, why not 'operator
> > bool() const;' if that is what you actually use it for?
Exactly for the reason you mention: A void * is convertable to a
bool, but not an integer (without cast). So this is safer.
> + if (lex)
> + vspace = VSpace(vsp);
>
> However, this isn't std::stream-like.
Of course it is.
27.4.5.3 basic_ios iostate flags functions [lib.iostate.flags]
operator void*() const
Returns:
If fail() then a null pointer; otherwise some non-null pointer to
indicate success.
Btw, I forgot to implement bool operator!() const.
> The std::stream-like interface would be
> if (lex.good())
> vspace = VSpace(vsp);
I think if (!lex.fail()) is the correct idiom, at least that's what is
required for the implementation of 'operator void *'. But I never
needed to understand these details as if (is) and if (!is) do what
is needed.
> Nonetheless, it does look a lot nicer. It also doesn't address the
> 'end_inset' thingy at all ;-)
Surprise.
Andre'
--
Those who desire to give up Freedom in order to gain Security, will not have,
nor do they deserve, either one. (T. Jefferson or B. Franklin or both...)
