On Mon, Mar 19, 2012 at 01:30:55AM +0000, Christos Zoulas wrote: > In article <20120318213646.ga13...@britannica.bec.de>, > Joerg Sonnenberger <jo...@britannica.bec.de> wrote: > > >That doesn't say anything about why it should be signed. It is just > >another place where signed is used for no good reason. > > Yes, it just explains that making unsigned is a larger impact change and > it would violate the standard.
...which is exactly why I asked why the standard requires it to be signed. It doesn't seem to make any sense to have this requirement. E.g. this is not a field returned by any system call, so it doesn't have to support -1 to flag error conditions. Joerg