Re: [Plone-i18n] i18n architecture question, ICU related

[Hanno S. <pl...@ha...>]

Michael Dexter wrote:
> 
> Does anyone know if consideration was given to using PyICU, the 
> python wrapper to ICU for use in Zope3 and in turn Plone?
> 
> http://pyicu.osafoundation.org/
> 
> Any insights as to why or why not?

The reason is quite simple. PyICU wasn't around at the time Zope's
i18n/locale support has been written ;) The i18n/l10n parts of Zope3 are
almost all about four years old now and havn't changed much in this time.

You might find some insights at
http://lists.osafoundation.org/pipermail/chandler-dev/2005-April/002662.html
for example where the OSAF was evaluating zope.locales as a candidate
for getting access to the ICU data. They turned this down, mainly
because zope.locales does not provide all of ICU but only the
information found in the XML files. If you look at
http://leilani.osafoundation.org/~bkirsch/i18n/ChandlerInternationalizationProposal.html
you find a lot of information how OSAF is using PyICU.

In the mid-term run I think it would be a good idea to look into
replacing Zope's own approach with PyICU, but this is quite a heavy task
and as long as there's nobody interested in it, it won't happen.

Right now the direction Zope3 takes is to use some different libraries
for the additional features ICU would provide as well. I think there's
both more advanced timezone support through pytz and I have seen some
collation (locale depend string sorting and comparison) work done by Jim
Fulton.

If you have a project (read time or money ;) that needs additional ICU
features I'm quite certain the Zope3 folks would be glad to see a
proposal for including PyICU into Zope3 (and somebody implementing it of
course ;).

Hanno