|
From: | Jerome Fisher |
Subject: | Re: [Monotone-devel] A better name for "context" |
Date: | Wed, 23 Jun 2004 15:02:13 +0200 |
User-agent: | Mozilla Thunderbird 0.7 (Windows/20040616) |
Nathaniel Smith wrote:
I really don't like "changeset", since to me it implies something like a recursive diff -- enough information to reconstruct a tree B given a tree A. The two problems being that "context"s don't exactly contain that detailed information, and that a "context" is an end-point, not an transition. Changesets describe transitions.
As I wrote, I don't like "changeset" for contexts either. But unless Graydon's implementation differs from what was proposed on this mailing list, I don't agree that contexts are more an end point than a transition. They describe both equally:
1) They describe the new state by specifying the new manifest ID.2) They describe the transition(s) from the ancestor(s) by explicitly listing all file additions, deletions, modifications and renames.
These all have the common problem that they assume a point of view; they're all talking about a "context" relative to its parents. Things quickly become incoherent when you want to say something like "What is this successor's parent successor's?"
Yeah, they're too directed. I felt that wasn't a good idea myself.
footstep, footprint, impress, trace, track, tread, rung, feat Hmm, "rung" has potential. Or a neologism, like "codestep" or something... or something like "mnode" for "monotone node"... or just "tone"?
Surely monotone can only have one tone? :)
[Prev in Thread] | Current Thread | [Next in Thread] |