Re: [PATCH INSTALLED]: const typedefs part 19/N

["Daniel Berlin" <dberlin at dberlin dot org>]

On 8/28/07, Dave Korn <dave.korn@artimi.com> wrote:
> On 28 August 2007 18:05, Daniel Berlin wrote:
>
> > On 8/28/07, Dave Korn <dave.korn@artimi.com> wrote:
> >> On 28 August 2007 17:10, Richard Guenther wrote:
> >>
> >>
> >>> Unless, of course, we start to use C++ language features.  Which is of course
> >>> the final goal of all this excercise.  But also the second.  The first
> >>> goal is to make
> >>> it easy to _try_ using C++ language features.
> >>
> >>   Yeah, I appreciate that.
> >>
> >>   OTOH that's somewhat further off down the road.
> >>
> >>   Also, when that time comes, we *do* have to have the discussion about
> >> increasing the requirements on ordinary users as opposed to developers.
> >
> > Yes, I can see how this would be a problem what with our *millions* of
> > users without C++ compilers.
>
>   Sarcasm is cheap, and falsified statistics are easy to pluck out of thin air,
> but they're both a bit insubstantial when it comes to establishing a point.

So wait, you really believe even >10% of our users don't have C++ compilers?
And that these people compile from source?

If you do, there is no point in arguing.  I would simply say you are
vastly out of touch with reality.

>   Is there any limit to the extra requirements we impose on end users beyond
> which you /would/ believe we're making it more difficult for them?

Honestly, if it makes it better for *developers*, and the requirements
are common, no.
The vast majority of our users do not compile from source.  The number
of people who make binary packages/compile from source/etc who do not
or could not easily get a C++ compiler is near 0.
Sorry.

What I see users begging us for is a faster, smaller, better code
producing compiler.  Not a compiler that "does not add requirements".