No one's expecting to be able to run a production site from HEAD at the moment I hope.
In fact I'm just about to put a site into production that's based on HEAD, last checkout 2003-03-07.
The main reason for using head is the cool i18n capability, makes more sense to use it rather than to hardcode translations in the .adp files (will contribute is_IS catalog files when I finish translating, or sooner if anyone wants what I already have).
The site does not use dotLRN, just plain OpenACS plus a custom package (which I'd like to contribute when I've generalized it a bit; it offers email subscriptions to RSS feeds and currently has a hardcoded interface in Icelandic, a bit similar to Simon's News Aggregator).
I know HEAD is not as safe as a numbered release, but am I making a seriously bad descision?
After last merge there was some problem which I submitted to bugracker with a suggested fix. Things seem to be running nicely now and I don't think I'll merge with OpenACS HEAD again before going live. But will a non-acs-core hacker like me have trouble keeping up with head in the future?
Reading an entry in Peter Marklund's blog, Back on the Edge of OpenACS Development, made me a bit more confident going this way :)
