3 number and dot..

[Hrvoje Niksic]

Paul McNett <p at ulmcnett.com> writes:

> Chris Mellon has given you the best response: use the locale module
> for this.

It may be the best choice as far as reuse is concerned, but it's not
without serious drawbacks.

For one, the locale model doesn't really allow forcing of the
separators.  Some locales, such as the "C" locale, define no thousand
separators whatsoever.  Since many Unix installations are set up to
use the "C" locale because it is the OS default, this problem must be
addressed.  Chris addresse it by forcing the locale to "US", but that
is a step away from the locale model because it consciously overrides
the user's locale preferences.  By doing that, one forces a particular
way of grouping thousands, without the possibility to using a
different grouping characters or to group by a different number of
digits (some locales group by tens of thousands).  That is quite
similar to what you get when you implement the desired grouping
yourself.

Setting the locale makes the code platform-dependent because different
platforms have different locale names.  For example, Chris's code
fails for me with "unsupported locale name" -- apparently, my system
calls the US locale is "en_US.utf8".  (Even specifying "en_US" doesn't
work.  It might be a Python or system problem, but it just doesn't
work.).  Finally, it makes the code OS-installation-dependent -- even
under the same OS, different installs can and do set up different
locales.

If his number presentation calls for thousand separators, coding them
manually is not an unreasonable implementation choice.

> It seems like you are looking for a spoonfed solution rather than
> looking for guidance on how to solve the problem for yourself. If
> I'm wrong about that assessment, I apologize in advance.

No argument here.