On Thu, Jun 30, 2011 at 11:59 AM, Christoph Nodes <[email protected]> wrote: > On Thu, 30 Jun 2011 11:21:02 +0200 > Bastien ROUCARIES <[email protected]> wrote: > >> With your patch even exact double like 1.0e0 are rounded up ! > > Right, thats exactly the problem we have with the our printf and > imagemagick: we always print the next greater representable double, > which in case of 1.0e0 and %.20g is ~1.000000000000000222. > > Surely, it would be smarter to only do this for "critical" numbers like > ~0.354999, for which we ASSUME that actually 0.355 is meant, but how > would you do that? There are no means to distinguish between 0.355 and > ~0.354999 because they both are represented by the same floating-point > number.
The patch on glibc bugzilla will do that, when truncating it will print that you want Bastien > > > Best regards, > Christoph > -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
