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
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
//! The [`Panel`] trait: the seam every dashboard widget goes through.
//!
//! **This is an in-tree seam, not a plugin API.** mirador is a binary crate
//! with no library target, so nothing outside it can name this trait; and even
//! if it could, `widgets::build` dispatches on a fixed `match` over
//! `WIDGET_NAMES` and each widget's settings are a field on `Config`. Adding a
//! widget is a pull request, and `CONTRIBUTING.md` lists the five places to
//! touch. A real plugin story needs three things this does not have: a runtime
//! registry instead of that `match`, per-widget config carried as an
//! unparsed `toml::Value` so a widget can own its own schema, and a library
//! target with a deliberate public surface. Adding only the last of those
//! would make the trait nameable without making it usable, which is worse than
//! the honest version.
//!
//! What the seam does buy, in-tree, is real: a panel owns its own state and
//! refresh cadence. The application shell is
//! responsible only for layout, focus, frames and dispatching ticks and key
//! events; it knows nothing about what any individual panel displays.
//!
//! # Why the description is six methods and not one
//!
//! `title`, `counter`, `bindings`, `max_width`, `max_height` and
//! `refresh_interval` are all "describe yourself", and folding them into one
//! `describe() -> PanelInfo` would take the trait from thirteen methods to
//! eight. It would also be slower and less useful, because the six are asked
//! for at different times and at very different rates: `max_width` and
//! `max_height` during layout, `refresh_interval` on the tick, `bindings` in
//! three separate places, and `title` and `counter` on **every frame** — the
//! last two from the shell's render loop, where nothing guards them.
//!
//! A single `describe()` would compute all six whenever any one was wanted,
//! and return an owned struct carrying a `String` title and an
//! `Option<String>` counter — so the cheapest question would allocate twice,
//! sixty times a second, per panel. The trait is wide because the shell's
//! questions are genuinely separate.
use Duration;
use Frame;
use ;
use Rect;
use crateBinding;
use crate;
/// Whether a panel handled an input event, or wants it to fall through to the
/// application's global bindings.
///
/// Shared by [`Panel::handle_key`] and [`Panel::handle_mouse`]: the two differ
/// in what they receive, not in how the shell reacts to the answer.
/// Everything a panel needs at draw time that it does not own itself.
/// Something that will get worse if it is ignored.
///
/// Deliberately not "something is wrong". Panels already show what is wrong in
/// their own frames — an overdue task is red, a stale reading says its age, a
/// failing fetch says so — and eight of the twelve do it today. Repeating all
/// of that in one line would produce a signal that is lit permanently, and a
/// permanently lit alarm is furniture: you stop seeing it inside a week, which
/// is the unread-badge failure reached from a different direction.
///
/// The bar is therefore **will this deteriorate if nobody acts in the next few
/// minutes**. A save that is failing loses work. An event about to start is
/// missed. A layout that did not persist is gone at the next launch. An overdue
/// task from three days ago is none of those: it is notable, it is already red,
/// and it will be exactly as overdue in an hour.
/// How bad, for picking one alert out of several.
///
/// Two levels because two is all the distinction that is needed: something is
/// about to happen, or something has broken. More would be a taxonomy nobody
/// consults.
/// A single dashboard widget.
///
/// Implementors are stored as `Box<dyn Panel>`, so the trait is deliberately
/// object safe.
/// Render a duration the way a person would say it.
///
/// Used wherever a panel shows data it fetched rather than computed, which is
/// the only honest way to present a reading that may have stopped updating.