[metapost] MetaPost 1.901 announcement

[Taco Hoekwater]

Hi,

On 10 Apr 2014, at 08:53, Franck Pastor <franck.pastor at mac.com> wrote:

> 
> By the way, has it been possible  to do — or plan — something about the units accuracy in double numbersystem mode mentioned lately by myself, and the arclength tolerances function problem mentioned by Harmut Henkel?

As I wrote, one big reason why I am quitting is that the remaining problems are well above my level of understanding.

For example, the comments in the web file on the arclength stuff imply that it is indeed using Simpson’s rule. However, it appears that the actual implementation is very much tailored to the ‘scaled’ method (probably to prevent all sorts of numerical overflows), and to confuse matters even more, a single combined helper function is used for both arclength and arctime, where a variable set to a fake ‘infinity’ value is used to differentiate the use cases. It just becomes too much for me … 

Best wishes,
Taco


(part of the problem is likely that I am not eighteen any more and my brain is not as flexible any more).