epics-ca-rs 0.18.2

EPICS Channel Access protocol client and server
Documentation
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
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
//! Helpers shared across the `caget` / `caput` / `cainfo` / `camonitor`
//! command-line binaries.

use epics_base_rs::types::EpicsValue;

/// Default CA CLI timeout in seconds when neither `-w` nor a usable
/// `EPICS_CLI_TIMEOUT` env var is set.
pub const DEFAULT_CLI_TIMEOUT_SECS: f64 = 1.0;

/// Read `EPICS_CLI_TIMEOUT` from the environment, falling back to
/// [`DEFAULT_CLI_TIMEOUT_SECS`] when unset, unparsable, or
/// non-finite/non-positive. Mirrors C `tool_lib.c:use_ca_timeout_env`
/// (commit 1d056c6) — the env var is consulted only when the caller
/// did not pass `-w`/`--wait` on the command line.
pub fn env_default_timeout() -> f64 {
    std::env::var("EPICS_CLI_TIMEOUT")
        .ok()
        .and_then(|s| s.parse::<f64>().ok())
        .filter(|v| v.is_finite() && *v > 0.0)
        .unwrap_or(DEFAULT_CLI_TIMEOUT_SECS)
}

/// Convert a user-supplied timeout (CLI `-w` or env var) into a
/// `std::time::Duration`. `Duration::from_secs_f64` panics on NaN /
/// infinity / negative values; clap accepts those literally so this
/// guard clamps to [`DEFAULT_CLI_TIMEOUT_SECS`] on the bad inputs.
/// Mirrors the spirit of epics-base 1655d68e (defensive against
/// pathological floats in CA timeout computation).
pub fn timeout_duration(secs: f64) -> std::time::Duration {
    let s = if secs.is_finite() && secs > 0.0 {
        secs
    } else {
        DEFAULT_CLI_TIMEOUT_SECS
    };
    std::time::Duration::from_secs_f64(s)
}

/// Field width the C tools (`caget` / `camonitor` / `caput -l`) use
/// when printing the PV name column: `printf("%-30s ...", name)` —
/// 30 chars left-aligned, then one space before the value. Mirrors
/// `epics-base/modules/ca/src/tools/tool_lib.c::print_value`'s width.
pub const PV_NAME_WIDTH: usize = 30;

/// Float number representation requested via `-e` / `-f` / `-g`.
#[derive(Debug, Clone, Copy)]
pub enum FloatStyle {
    /// `%g` — shortest of `%f` / `%e`. C tools default. Precision is
    /// the count of *significant* digits.
    G,
    /// `%e` — scientific notation. Precision is digits after decimal.
    E,
    /// `%f` — fixed-point. Precision is digits after decimal.
    F,
}

/// Float formatting options. C precision defaults to 6 for all three
/// styles per `printf(3)`.
#[derive(Debug, Clone, Copy)]
pub struct FloatFormat {
    pub style: FloatStyle,
    pub precision: u32,
}

impl Default for FloatFormat {
    fn default() -> Self {
        Self {
            style: FloatStyle::G,
            precision: 6,
        }
    }
}

/// Integer formatting requested via `-0x` / `-0o` / `-0b` (base for
/// integer types) and `-lx` / `-lo` / `-lb` (round-float-to-long
/// then print in base). C tool default is decimal.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum IntStyle {
    Dec,
    Hex,
    Oct,
    Bin,
}

/// Per-tool CLI formatting state.
#[derive(Debug, Clone)]
pub struct ValueFormat {
    pub float: FloatFormat,
    pub int_style: IntStyle,
    /// Round-float-to-long-and-render in `int_style` (the `-lx` / `-lo`
    /// / `-lb` C-tool flags). Only applies to floating-point values.
    pub float_as_int: bool,
    /// `-n` flag: print enum value as its integer index instead of
    /// the menu string.
    pub enum_as_number: bool,
    /// `-S` flag: render `DBR_CHAR` arrays as a NUL-terminated string
    /// (long-string CA convention).
    pub char_array_as_string: bool,
    /// `-# <count>` flag: cap displayed array elements; `None` means
    /// "all". Acts on display only — the request still asks for the
    /// requested count from the IOC.
    pub max_elements: Option<usize>,
    /// `-F <ofs>` flag: replacement field separator. Defaults to a
    /// single space.
    pub field_separator: char,
}

impl Default for ValueFormat {
    fn default() -> Self {
        Self {
            float: FloatFormat::default(),
            int_style: IntStyle::Dec,
            float_as_int: false,
            enum_as_number: false,
            char_array_as_string: false,
            max_elements: None,
            field_separator: ' ',
        }
    }
}

/// Render `EpicsValue` for CA tool output, matching C `tool_lib.c::
/// print_value`. Scalars are bare (no count prefix); arrays are
/// `count<sep>v0<sep>v1...` (the count is part of the value, not the
/// PV-name column). Enum strings are NOT resolved here — caller passes
/// `enum_strings = Some(&["off","on",...])` when it has them, else
/// the integer index is used (matches `-n` flag default when no enum
/// metadata is available). `format_value` does not emit a trailing
/// newline.
///
/// `req_elems_present` mirrors C `caget.c:286` / the `PRN_TIME_VAL_STS`
/// macro (`tool_lib.c:486`): the array element-count prefix is emitted
/// only when `reqElems || pv->nElems > 1`. Pass `true` when the user
/// supplied `-#` on the command line; a genuine 1-element waveform read
/// without `-#` then prints just the value with no count prefix.
pub fn format_value(
    v: &EpicsValue,
    fmt: &ValueFormat,
    enum_strings: Option<&[String]>,
    req_elems_present: bool,
) -> String {
    let sep = fmt.field_separator;
    match v {
        EpicsValue::String(s) => s.clone(),
        EpicsValue::Short(n) => format_int_i64(*n as i64, fmt.int_style),
        EpicsValue::Long(n) => format_int_i64(*n as i64, fmt.int_style),
        EpicsValue::Int64(n) => format_int_i64(*n, fmt.int_style),
        EpicsValue::UInt64(n) => format_int_u64(*n, fmt.int_style),
        EpicsValue::Char(n) => format_int_i64((*n as i8) as i64, fmt.int_style),
        EpicsValue::Enum(idx) => format_enum(*idx as i64, fmt, enum_strings),
        EpicsValue::Float(x) => format_float(*x as f64, fmt),
        EpicsValue::Double(x) => format_float(*x, fmt),
        EpicsValue::ShortArray(arr) => render_array_int(
            arr.iter().map(|&n| n as i64),
            arr.len(),
            fmt,
            sep,
            req_elems_present,
        ),
        EpicsValue::LongArray(arr) => render_array_int(
            arr.iter().map(|&n| n as i64),
            arr.len(),
            fmt,
            sep,
            req_elems_present,
        ),
        EpicsValue::Int64Array(arr) => {
            render_array_int(arr.iter().copied(), arr.len(), fmt, sep, req_elems_present)
        }
        EpicsValue::UInt64Array(arr) => render_array_iter(
            arr.iter().map(|&n| format_int_u64(n, fmt.int_style)),
            arr.len(),
            fmt,
            sep,
            req_elems_present,
        ),
        EpicsValue::EnumArray(arr) => {
            let mut parts = Vec::with_capacity(arr.len() + 1);
            if req_elems_present || arr.len() > 1 {
                parts.push(arr.len().to_string());
            }
            let take = fmt.max_elements.unwrap_or(arr.len()).min(arr.len());
            for &idx in &arr[..take] {
                parts.push(format_enum(idx as i64, fmt, enum_strings));
            }
            parts.join(&sep.to_string())
        }
        EpicsValue::FloatArray(arr) => render_array_iter(
            arr.iter().map(|&x| format_float(x as f64, fmt)),
            arr.len(),
            fmt,
            sep,
            req_elems_present,
        ),
        EpicsValue::DoubleArray(arr) => render_array_iter(
            arr.iter().map(|&x| format_float(x, fmt)),
            arr.len(),
            fmt,
            sep,
            req_elems_present,
        ),
        EpicsValue::CharArray(arr) => {
            // C `caget.c` renders a CHAR array as a long-string only when
            // `charArrAsStr && (reqElems || nElems > 1)` — a 1-element
            // CHAR array with `-S` but no `-#` falls through to numeric.
            if fmt.char_array_as_string && (req_elems_present || arr.len() > 1) {
                // Long-string convention: bytes up to first NUL.
                let end = arr.iter().position(|&b| b == 0).unwrap_or(arr.len());
                String::from_utf8_lossy(&arr[..end]).into_owned()
            } else {
                render_array_int(
                    arr.iter().map(|&b| (b as i8) as i64),
                    arr.len(),
                    fmt,
                    sep,
                    req_elems_present,
                )
            }
        }
        EpicsValue::StringArray(arr) => {
            let mut parts = Vec::with_capacity(arr.len() + 1);
            if req_elems_present || arr.len() > 1 {
                parts.push(arr.len().to_string());
            }
            let take = fmt.max_elements.unwrap_or(arr.len()).min(arr.len());
            parts.extend(arr[..take].iter().cloned());
            parts.join(&sep.to_string())
        }
    }
}

fn render_array_int<I: Iterator<Item = i64>>(
    iter: I,
    total: usize,
    fmt: &ValueFormat,
    sep: char,
    req_elems_present: bool,
) -> String {
    let take = fmt.max_elements.unwrap_or(total).min(total);
    let mut parts = Vec::with_capacity(take + 1);
    if req_elems_present || total > 1 {
        parts.push(total.to_string());
    }
    for n in iter.take(take) {
        parts.push(format_int_i64(n, fmt.int_style));
    }
    parts.join(&sep.to_string())
}

fn render_array_iter<I: Iterator<Item = String>>(
    iter: I,
    total: usize,
    fmt: &ValueFormat,
    sep: char,
    req_elems_present: bool,
) -> String {
    let take = fmt.max_elements.unwrap_or(total).min(total);
    let mut parts = Vec::with_capacity(take + 1);
    if req_elems_present || total > 1 {
        parts.push(total.to_string());
    }
    parts.extend(iter.take(take));
    parts.join(&sep.to_string())
}

fn format_enum(idx: i64, fmt: &ValueFormat, enum_strings: Option<&[String]>) -> String {
    if !fmt.enum_as_number
        && let Some(strs) = enum_strings
        && idx >= 0
        && (idx as usize) < strs.len()
    {
        return strs[idx as usize].clone();
    }
    format_int_i64(idx, fmt.int_style)
}

fn format_int_i64(n: i64, style: IntStyle) -> String {
    match style {
        IntStyle::Dec => n.to_string(),
        // C `%lx` / `%lo` / `%lb` print the bit pattern of a signed
        // integer reinterpreted as unsigned of the same width. We
        // emulate that with the `as u64` cast.
        IntStyle::Hex => format!("0x{:x}", n as u64),
        IntStyle::Oct => format!("0o{:o}", n as u64),
        IntStyle::Bin => format!("0b{:b}", n as u64),
    }
}

/// Format a `DBF_UINT64` value. Unlike `format_int_i64`, the decimal
/// rendering keeps the full unsigned range — a value above `i64::MAX`
/// must print as a positive integer, not a negative one.
fn format_int_u64(n: u64, style: IntStyle) -> String {
    match style {
        IntStyle::Dec => n.to_string(),
        IntStyle::Hex => format!("0x{n:x}"),
        IntStyle::Oct => format!("0o{n:o}"),
        IntStyle::Bin => format!("0b{n:b}"),
    }
}

fn format_float(x: f64, fmt: &ValueFormat) -> String {
    if fmt.float_as_int {
        // C tool: round-half-to-even via `lroundl`, then format as int.
        let rounded = if x.is_nan() { 0i64 } else { x.round() as i64 };
        return format_int_i64(rounded, fmt.int_style);
    }
    if !x.is_finite() {
        // C printf prints "nan" / "inf" / "-inf"; Rust matches by
        // default (`{}` on f64 yields the same lowercase forms).
        return format!("{x}");
    }
    let p = fmt.float.precision as usize;
    match fmt.float.style {
        FloatStyle::F => format!("{x:.p$}"),
        FloatStyle::E => format_e(x, p),
        FloatStyle::G => format_g(x, p.max(1)),
    }
}

/// Decimal exponent of `abs` after rounding to `precision` significant
/// digits — i.e. `floor(log10(round_to_sig_digits(abs, precision)))`.
///
/// C `%g` rounds to `precision` significant digits FIRST and only then
/// decides between `%e` and `%f`. At a rounding boundary the rounded
/// magnitude can tick up by a power of ten (e.g. `999999.5` at
/// precision 6 rounds to `1000000`, exponent 5 → 6), which flips the
/// fixed-vs-scientific choice. Computing the decision exponent from the
/// UNROUNDED value misses that — see the regression test
/// `g_rounding_boundary_picks_scientific`.
fn decision_exponent(abs: f64, precision: usize) -> i32 {
    let raw_exp = abs.log10().floor() as i32;
    // Scale so the value has `precision` digits before the decimal
    // point, round half-to-even, and read back the magnitude. If the
    // round carries into a new decade the exponent increments.
    let scale = 10f64.powi(precision as i32 - 1 - raw_exp);
    // For magnitudes near the f64 range limits the scale factor can
    // overflow to ±inf (or underflow to 0); `abs * scale` then yields a
    // non-finite product whose `log10` saturates to a garbage exponent.
    // The rounded magnitude cannot meaningfully differ from the raw one
    // at those scales, so fall back to `raw_exp`.
    if !scale.is_finite() || scale == 0.0 {
        return raw_exp;
    }
    let rounded_scaled = (abs * scale).round();
    if !rounded_scaled.is_finite() || rounded_scaled <= 0.0 {
        return raw_exp;
    }
    raw_exp + (rounded_scaled.log10().floor() as i32 - (precision as i32 - 1))
}

/// `%g`-equivalent formatter. C semantics: choose `%e` or `%f`
/// depending on the exponent, drop trailing zeros and the trailing
/// decimal point. Precision is the *significant-digit* count.
fn format_g(x: f64, precision: usize) -> String {
    if x == 0.0 {
        return "0".to_string();
    }
    let abs = x.abs();
    // C `%g` rounds to `precision` significant digits before choosing
    // the format, so the decision exponent must come from the rounded
    // magnitude, not the raw value.
    let exp = decision_exponent(abs, precision);
    // C `%g` uses fixed-point when `precision > exp >= -4`. Compare as
    // i32 to avoid the silent `i32 → usize` wrap for negative `exp`.
    if exp >= -4 && exp < precision as i32 {
        // Fixed-point. Digits after the decimal point = precision-1-exp.
        let digits = (precision as i32 - 1 - exp).max(0) as usize;
        let s = format!("{x:.digits$}");
        trim_g_fixed(&s)
    } else {
        format_g_scientific(x, precision)
    }
}

fn trim_g_fixed(s: &str) -> String {
    if !s.contains('.') {
        return s.to_string();
    }
    s.trim_end_matches('0').trim_end_matches('.').to_string()
}

fn format_g_scientific(x: f64, precision: usize) -> String {
    // Rust `{:e}` precision means digits after the decimal in the
    // mantissa. C `%g` with N significant digits is mantissa
    // precision N-1.
    let s = format!("{:.*e}", precision - 1, x);
    rewrite_rust_e_as_c(&s, true)
}

/// Pure `%e`: precision is digits AFTER the decimal point.
fn format_e(x: f64, precision: usize) -> String {
    let s = format!("{x:.precision$e}");
    rewrite_rust_e_as_c(&s, false)
}

/// Rust formats scientific as `1.23e2`; C as `1.23e+02` (signed,
/// 2-digit exponent minimum). Optionally also strips the trailing
/// zeros from the mantissa (the `%g` post-trim behaviour).
fn rewrite_rust_e_as_c(s: &str, trim_mantissa: bool) -> String {
    let Some(e_pos) = s.find('e') else {
        return s.to_string();
    };
    let mantissa = &s[..e_pos];
    let exp_part = &s[e_pos + 1..];
    let mantissa_out = if trim_mantissa && mantissa.contains('.') {
        let t = mantissa.trim_end_matches('0').trim_end_matches('.');
        t.to_string()
    } else {
        mantissa.to_string()
    };
    let (sign, digits) = if let Some(d) = exp_part.strip_prefix('-') {
        ('-', d)
    } else if let Some(d) = exp_part.strip_prefix('+') {
        ('+', d)
    } else {
        ('+', exp_part)
    };
    let exp_padded = if digits.len() < 2 {
        format!("{sign}0{digits}")
    } else {
        format!("{sign}{digits}")
    };
    format!("{mantissa_out}e{exp_padded}")
}

#[cfg(test)]
mod tests {
    use super::*;

    fn fmt_default() -> ValueFormat {
        ValueFormat::default()
    }

    #[test]
    fn g_default_precision_matches_c() {
        // C `printf("%g", 475.123)` → "475.123"
        assert_eq!(format_g(475.123, 6), "475.123");
        // C `printf("%g", 1.0)` → "1"
        assert_eq!(format_g(1.0, 6), "1");
        // C `printf("%g", 0.0)` → "0"
        assert_eq!(format_g(0.0, 6), "0");
        // C `printf("%g", 1e-5)` → "1e-05"
        assert_eq!(format_g(1e-5, 6), "1e-05");
        // C `printf("%g", 1e10)` → "1e+10"
        assert_eq!(format_g(1e10, 6), "1e+10");
        // C `printf("%g", 0.0001)` → "0.0001" (boundary)
        assert_eq!(format_g(0.0001, 6), "0.0001");
        // C `printf("%g", 1234567.0)` → "1.23457e+06"
        assert_eq!(format_g(1234567.0, 6), "1.23457e+06");
    }

    /// C `%g` rounds to `precision` significant digits BEFORE deciding
    /// between `%e` and `%f`. At a rounding boundary the rounded
    /// magnitude can carry into a new decade and flip the choice:
    /// `printf("%g", 999999.5)` → "1e+06" (not "1000000"), because the
    /// rounded value `1000000` has exponent 6 >= precision 6.
    #[test]
    fn g_rounding_boundary_picks_scientific() {
        // 999999.5 rounds up to 1000000 → exponent ticks 5 → 6 → %e.
        assert_eq!(format_g(999999.5, 6), "1e+06");
        // Just below the boundary stays fixed-point.
        assert_eq!(format_g(999998.0, 6), "999998");
        // 9.999995 rounds to 10 (exponent 0 → 1, still fixed range).
        assert_eq!(format_g(9.999995, 6), "10");
        // Negative value at the same boundary keeps the sign.
        assert_eq!(format_g(-999999.5, 6), "-1e+06");
    }

    #[test]
    fn g_extreme_magnitudes_do_not_produce_garbage() {
        // Magnitudes near the f64 range limits make the internal scale
        // factor overflow/underflow; decision_exponent must fall back
        // to the raw exponent instead of saturating to a garbage value.
        // Tiny: classifies as scientific (exp < -4).
        assert_eq!(format_g(1e-308, 6), "1e-308");
        assert_eq!(format_g(5e-300, 6), "5e-300");
        // Huge: classifies as scientific (exp >= precision).
        assert_eq!(format_g(1e308, 6), "1e+308");
        // Smallest normal f64 — no panic, scientific form.
        assert!(format_g(f64::MIN_POSITIVE, 6).contains("e-"));
    }

    #[test]
    fn e_format_matches_c() {
        // C `printf("%e", 1.5)` → "1.500000e+00"
        assert_eq!(format_e(1.5, 6), "1.500000e+00");
        // C `printf("%.2e", 1234.5)` → "1.23e+03"
        assert_eq!(format_e(1234.5, 2), "1.23e+03");
    }

    #[test]
    fn negative_int_hex_matches_c() {
        // C `printf("%lx", (long)-1)` → "ffffffffffffffff"
        assert_eq!(format_int_i64(-1, IntStyle::Hex), "0xffffffffffffffff");
    }

    #[test]
    fn array_renders_count_then_values() {
        let v = EpicsValue::DoubleArray(vec![1.0, 2.5, 3.0]);
        let s = format_value(&v, &fmt_default(), None, false);
        // C: `3 1 2.5 3` (count + space-separated %g values)
        assert_eq!(s, "3 1 2.5 3");
    }

    /// C `caget.c:286` gates the count prefix on `reqElems || nElems > 1`.
    /// A genuine 1-element waveform read WITHOUT `-#` prints just the
    /// value, no `1 ` prefix.
    #[test]
    fn single_element_array_omits_count_without_req_elems() {
        let v = EpicsValue::DoubleArray(vec![2.5]);
        // No `-#` on the command line → no count prefix.
        assert_eq!(format_value(&v, &fmt_default(), None, false), "2.5");
        // `-#` supplied → count prefix returns even for 1 element.
        assert_eq!(format_value(&v, &fmt_default(), None, true), "1 2.5");
        // Multi-element always carries the count prefix.
        let v2 = EpicsValue::DoubleArray(vec![1.0, 2.5]);
        assert_eq!(format_value(&v2, &fmt_default(), None, false), "2 1 2.5");
    }

    #[test]
    fn enum_with_strings_renders_string() {
        let strs = vec!["off".to_string(), "on".to_string()];
        let v = EpicsValue::Enum(1);
        let s = format_value(&v, &fmt_default(), Some(&strs), false);
        assert_eq!(s, "on");
    }

    #[test]
    fn enum_n_flag_renders_index() {
        let strs = vec!["off".to_string(), "on".to_string()];
        let v = EpicsValue::Enum(1);
        let mut fmt = fmt_default();
        fmt.enum_as_number = true;
        let s = format_value(&v, &fmt, Some(&strs), false);
        assert_eq!(s, "1");
    }

    #[test]
    fn char_array_long_string_strips_at_nul() {
        let v = EpicsValue::CharArray(b"hello\0xxxx".to_vec());
        let mut fmt = fmt_default();
        fmt.char_array_as_string = true;
        assert_eq!(format_value(&v, &fmt, None, false), "hello");
    }

    #[test]
    fn float_as_int_rounds_then_renders() {
        let v = EpicsValue::Double(1234.6);
        let mut fmt = fmt_default();
        fmt.float_as_int = true;
        fmt.int_style = IntStyle::Hex;
        // 1235 = 0x4d3
        assert_eq!(format_value(&v, &fmt, None, false), "0x4d3");
    }

    #[test]
    fn pv_name_width_constant_is_30() {
        // Lock the pad width so a future tweak gets caught.
        assert_eq!(PV_NAME_WIDTH, 30);
    }

    /// `Duration::from_secs_f64` panics on NaN / +Inf / negative.
    /// `timeout_duration` must clamp those to the safe default rather
    /// than panic — the user-supplied value reaches us via clap which
    /// happily parses "NaN", "inf", "-1" as f64.
    #[test]
    fn timeout_duration_clamps_pathological_floats() {
        let d = timeout_duration(f64::NAN);
        assert_eq!(d.as_secs_f64(), DEFAULT_CLI_TIMEOUT_SECS);
        let d = timeout_duration(f64::INFINITY);
        assert_eq!(d.as_secs_f64(), DEFAULT_CLI_TIMEOUT_SECS);
        let d = timeout_duration(f64::NEG_INFINITY);
        assert_eq!(d.as_secs_f64(), DEFAULT_CLI_TIMEOUT_SECS);
        let d = timeout_duration(-1.0);
        assert_eq!(d.as_secs_f64(), DEFAULT_CLI_TIMEOUT_SECS);
        let d = timeout_duration(0.0);
        assert_eq!(d.as_secs_f64(), DEFAULT_CLI_TIMEOUT_SECS);
    }

    /// Sane positive values pass through unchanged.
    #[test]
    fn timeout_duration_preserves_positive_finite() {
        let d = timeout_duration(2.5);
        assert!((d.as_secs_f64() - 2.5).abs() < 1e-9);
    }

    /// Env-var path must reject the same pathological set so a
    /// misconfigured `EPICS_CLI_TIMEOUT` doesn't propagate. Serialised
    /// because env-var mutation races every other test that consults
    /// `EPICS_CLI_TIMEOUT` (none today, but #[serial] is cheap insurance).
    #[serial_test::serial]
    #[test]
    fn env_default_timeout_rejects_nan_inf() {
        // SAFETY: serial_test::serial guarantees no concurrent env access.
        unsafe { std::env::set_var("EPICS_CLI_TIMEOUT", "NaN") };
        assert_eq!(env_default_timeout(), DEFAULT_CLI_TIMEOUT_SECS);
        unsafe { std::env::set_var("EPICS_CLI_TIMEOUT", "inf") };
        assert_eq!(env_default_timeout(), DEFAULT_CLI_TIMEOUT_SECS);
        unsafe { std::env::set_var("EPICS_CLI_TIMEOUT", "-3") };
        assert_eq!(env_default_timeout(), DEFAULT_CLI_TIMEOUT_SECS);
        unsafe { std::env::set_var("EPICS_CLI_TIMEOUT", "2.5") };
        assert!((env_default_timeout() - 2.5).abs() < 1e-9);
        unsafe { std::env::remove_var("EPICS_CLI_TIMEOUT") };
    }

    #[test]
    fn max_elements_caps_array() {
        let v = EpicsValue::LongArray((0..10).collect());
        let mut fmt = fmt_default();
        fmt.max_elements = Some(3);
        // `-#` implies `req_elems_present` so the count prefix is present.
        let s = format_value(&v, &fmt, None, true);
        // Total count is full (10) per C `caget -# 3` behaviour:
        //   "10 0 1 2"
        assert_eq!(s, "10 0 1 2");
    }
}