Bug#632031: convert: gray colorspace creates broken postscript

[Christoph Nodes <nodes@bfw-online.de>]

On Thu, 30 Jun 2011 15:04:11 +0200
Bastien ROUCARIES <roucaries.bastien@gmail.com> wrote:

> I will close this bug 

Thanks for your effort. I already stated that the problem is completely
on our side and that its not caused by imagemagick, so there is no bug.

> if you could not reproduce the following
> faillure with:
> - printf("%.20f\n",someint < 2^64 -1)
> 
> and fesetround(mode)
> set to rounding up mode
> 
> I suspect it could not be the case
> Between 2^52=4,503,599,627,370,496 and 2^53=9,007,199,254,740,992 the
> representable numbers are exactly the integers. For the next range,
> from 253 to 254, everything is multiplied by 2, so the representable
> numbers are the even ones, etc. Conversely, for the previous range
> from 251 to 252, the spacing is 0.5, etc.
> 
> You are no conformant to IEEE 754-2008 or IEEE_754-1985

thats true.

> 0^- will be rounded to 0^+ with your shame
>
> The only bug is that imagemagick should test IEEE_754-1985

The problem was quite easily solved by not do our "rounding" for
other format strings than %f, so that the %g printing in imagemagick is
not affected. Additionally we excluded floating point numbers with zero
fractional part, so even with %.20f integers are printed correctly.


Best regards,
Christoph