Internationalisation pipeline

[mark-lester]

I believe the translation substitution process in a templating architecture should be performed on demand upon an initial request, and ideally affected templates be marked for reprint on specific locales upon dictionary updates.
The whole internationalisation business is cornered by proprietary dictionaries. There is no free way to do that for simple/common translations. It would be nice to liberate that and allow everyone to keep their funky languages. We can even go as far as crowd sourcing the userbase for translations.
I wrote extensions to Liquid to handle two phase server side execution and inline . I also ripped up gettext. It's the standard transport and is sufficient for that, but it requires/assumes in store execution which if we are going to do arbitrary content across arbitrary locales is obviously insufficient. It's an awful thing to aggregate.
I don't know if this is of interest, I saw pre generation architecture and Node but haven't dug deeper.