Antoine Pitrou wrote:
> The problem seems to be that the thousands separator on the Solaris
> variant of en_US is an empty string (rather than a comma) (*), and
> apparently it hits a bug in the padding mechanism (which perhaps assumes
> that the thousands separator is always a 1-character string).
> 
> (*) (another reason not to use any C locale-based mechanism for
> localization, by the way...)

I've come to believe this, too. I'm working on cleaning up the C 
implementations so I can do all of the locale-based formating without 
using the locale functions. I'll use the localeconv values, but that's it.