Reformatted excerpts from William Morgan's message of 2008-10-02: > Give me a few more days to clean it up, and I'll throw up a git branch > somewhere. Definitely keep bugging me.
Status update: after trying a couple things over the past few months, I think I have settled on a workable architecture. It's split into three components: a general searchable-tree-of-documents store, an email-specific layer that handles things like attachments and the JWZ threading algorithm, and, of course, a client. I have a partial implementation of the first layer, which I'm currently working on converting from an earlier architecture into the current one. I'm making both the base document store, and the search index, pluggable, and have a simple implementation of each: the document store just keeps documents as files on your local disk, and the search component uses Sphinx. My current plan is to put up a git repo this weekend of what I have, so that others can start hacking on it too. It will basically be the first layer as I described above, though without any of the tree stuff actually working. It will use Thrift as the interface layer, so writing clients should be possible in any language. -- William <[EMAIL PROTECTED]> _______________________________________________ sup-talk mailing list [email protected] http://rubyforge.org/mailman/listinfo/sup-talk
