Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: OT: The joys of email (was: Re: CONLANG/ZBB crossover)

From:Tristan Plumb <lingua@...>
Date:Monday, May 14, 2007, 13:57
> (Quoted matter from T. A. McLeay, 2007.05.14 22:36:49 +1000)
> Ah, that's true, and in those circumstances I often think a bit about > which message should be the official reply (even tho the decision > isn't usually based on anything more than a whim/what came first). In > such circumstances, of course, you could simply reply to your original > message, and in that case you'd be no worse-off than in a normal > subject-based email forum.
Or indeed, it's perfectly legal to have multiple entries in In-Reply-To, even if most clients (and archive software) only understand one of them. That said, I don't know if Icedove (another Debian user?) lets you play with headers like that (or whichever client you attempted to switch to).
> Actually, I suspect it's probably more because I have *no* idea how > you'd visualise it concisely. A tree view can be done just with a > regular list with an indent, but a graph would probably need to be > full-on 2d. A lot of work for what's in practice (at least on email > forums like these) a relatively infrequent occurence.
I'm envisioning something like the view in gitk, there's a space filled with vertical lines, threads, and dots thereon, the individual messages. Messages with multiple replies have n lines forked at their dot, and an inclusive reply unifies these forked lines. Nodes with multiple parents are common in revision control systems, though here we have rather more diverging, and less unification, borrowing ideas there from makes sense.
> There's the less-used References header which lists all ancestors, > which an intelligent client could fill-in automatically when you > copy-and-paste from another email, so you wouldn't necessarily have to > work out where posts came from based on content.
Still, many clients don't do anything, and to reconstruct threads where these clients play part, some sort of probabilistic thing with contents and headers both is required. (I've been working, albeit slowly, on it!) (And then the case without headers or quoted text, manual intervention?) enjoy all, tristan -- All original matter is hereby placed immediately into the public domain.