Greg Hudson <ghudson@MIT.EDU> writes:
> Well, one of the cool things about the edit_fns concept from my
> standpoint is that producers and consumers can be plugged together
> more or less arbitrarily, and you can write shims which accept calls
> from an arbitrary producer and relay those calls (modified, perhaps)
> to an arbitrary consumer. For that to continue to be the case, you
> have to code producers to the least common denominator of the
> consumers.
Yeah. What is it about the Gregs on this project -- they're always
right (even when they disagree?).
Anyway, I think what you both say is wise. We shouldn't do away with
that arbitrary mix-and-match ability of editor producers and consumers
at this stage, until we have a more solid feel for how editors get
used.
Let's please go with the least common denominator for now, which (I
believe) is depth-first calls. The XML generator depends on that, the
working copy is a lot happier with it to, and other uses are (?) as
comfortable with depth-first as with any other order.
Greg (Stein), you mentioned that you do breadth first. Which editor
are you calling into for this, and for what purpose? I think if we
can compare a depth-first-required situation with your breadth-first
situation, we'll understand things a lot better.
Received on Sat Oct 21 14:36:09 2006