1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
//! **The records pane's reads** (§4.29, bl-2cf7): what the selected
//! conversation's loop did, what its worktree holds, which config governs it
//! and what sits in its inbox — asked only while somebody is looking. Split
//! from [`super::workspace`] at the design-time budget (bl-1d79), when the
//! spine left this fan for the selection's own (§4.47 item 3).
use Path;
use aimed;
use about;
use crateLink;
use crateChannel;
/// **The records pane's six standing reads**, asked only while it is open.
///
/// A body of its own rather than six more calls inside [`ask`]: that function
/// is *the questions that name one workspace*, and this is one pane's — the
/// pane with more reads under it than any other, and the one three balls grew.
/// The pair bl-2cf7 landed is what the loop did and what its worktree holds;
/// bl-b52c's spine is the two whose answer the `fork` control's one argument
/// comes off, because a control offered on a notch this seat has not been
/// answered about would be a control with nothing to carry; and bl-3257's two
/// were the conversation's own row and its undelivered mail, of which the row
/// went back out to the conversation's own standing set at bl-7b03 — this pane
/// still reads it, off the same field, but the header over the TRANSCRIPT
/// needs it whether or not this pane is open. The sixth is the workspace's lineage
/// listing (bl-ed20), the config pane's read asked here too, because the names
/// it answers are the only arguments the workflow mark's control can carry
/// (`crate::ui::records::workflow`) — the spine's reason, one read over. It
/// lands on the one `Model::lineages` either pane fills.
///
/// **The pane's seventh read is not here and never will be**: `step` is
/// addressed at one row of the ledger rather than at the pane, so the control
/// on that row posts it (`crate::ui::model::deep`).
pub