Re: [RFC] introducing new locale-db implementation (Re: lib/39662: shortcomings in LC_{MONETARY,NUMERIC,TIME,MESSAGES} db format)

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Fri, Jan 02, 2009 at 01:13:41PM +0900, Takehiko NOZAKI wrote:
> however the main purpose of introducing this locale caching mechanism is
> not for performance issue, but for multi-locale implementation.

I'd strongly prefer if you introduce a proper multi-locale interface and
*not* a thread-locale interface. A thread-locale interface can easily
mapped to a correct multi-locale interface, but not the reserve. E.g.
the multi-locale interface might add an argument to all locale-specific
function (or provide an alternative hook with such an interface). This
might be too late for netbsd-5, but it certainly would be a nice item
for the merge of the time_t branch. Basically, all the ctype functions
wuold implicitly refer to the default locale and a separate interface be
provided  for function/macro + locale handle.

Joerg

P.S.: thread-locale interface can be easily mapped using a thread-locale
variable, both for multi- and single-threaded applications. The more
interesting case to support is distuinguishing between different locales
in one thread.