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
//!
//! Definition of histograms.
//!
//! Histograms represent measures on a set of entities known at
//! compile-time. In some cases, the set of entities is not known
//! at compile-time (e.g. plug-ins, dates), in which case you should
//! rather use [Keyed histograms](../keyed.index.html).
//!

use rustc_serialize::json::Json;

use std::marker::PhantomData;
use std::mem::size_of;
use std::sync::atomic::{AtomicBool, Ordering};

use misc::{Flatten, LinearBuckets, SerializationFormat, vec_resize, vec_with_size};
use task::{BackEnd, Op, PlainRawStorage};
use service::{Service, PrivateAccess};
use indexing::*;

///
/// A plain histogram.
///
/// Histograms do not implement `Sync`, so an instance of `Histogram`
/// cannot be shared by several threads. However, any histogram can be
/// cloned as needed for concurrent use.
///
/// # Performance
////
/// Cloning a histogram is relatively cheap, both in terms of memory
/// and in terms of speed (most histograms weigh ~40bytes on a x86-64
/// architecture).
///
/// When the telemetry service is inactive, recording data to a
/// histogram is very fast (essentially a dereference and an atomic
/// fetch). When the telemetry service is active, the duration of
/// recording data is comparable to the duration of sending a simple
/// message to a `Sender`.
///
pub trait Histogram<T> : Clone {
    ///
    /// Record a value in this histogram.
    ///
    /// If the service is currently inactive, this is a noop.
    ///
    fn record(&self, value: T) {
        self.record_cb(|| Some(value))
    }

    ///
    /// Record a value in this histogram, as provided by a callback.
    ///
    /// If the service is currently inactive, this is a noop.
    ///
    /// If the callback returns `None`, no value is recorded.
    ///
    fn record_cb<F>(&self, _: F) where F: FnOnce() -> Option<T>;
}


/// Back-end features specific to plain histograms.
impl BackEnd<Plain> {
    /// Instruct the Telemetry Task to record a value in an
    /// already registered histogram.
    fn raw_record(&self, k: &Key<Plain>, value: u32) {
        self.sender.send(Op::RecordPlain(k.index, value)).unwrap();
    }

    /// Instruct the Telemetry Task to record the result of a callback
    /// in an already registered histogram.
    fn raw_record_cb<F, T>(&self, cb: F) -> bool where F: FnOnce() -> Option<T>, T: Flatten {
        if let Some(k) = self.get_key() {
            if let Some(v) = cb() {
                self.raw_record(&k, v.as_u32());
                true
            } else {
                false
            }
        } else {
            false
        }
    }
}

///
/// A histogram that ignores any input.
///
/// Useful for histograms that can be activated/deactivated either at
/// compile-time (e.g. because they are attached to specific versions
/// of the application) or during startup (e.g. depending on
/// command-line options).
///
pub struct Ignoring<T> {
    witness: PhantomData<T>,
}

impl<T> Ignoring<T> {
    ///
    /// Create an histogram that ignores any input.
    ///
    /// `Ignoring` histograms are effectively implemented as empty
    /// structs, without a back-end, so they take no memory.
    ///
    pub fn new() -> Ignoring<T> {
        Ignoring {
            witness: PhantomData
        }
    }
}

impl<T> Histogram<T> for Ignoring<T> {
    fn record_cb<F>(&self, _: F) where F: FnOnce() -> Option<T>  {
        return;
    }
}

impl<T> Clone for Ignoring<T> {
    fn clone(&self) -> Self {
        Ignoring {
            witness: PhantomData
        }
    }
}

///
///
/// Flag histograms.
///
/// This histogram has only two states. Until the first call to
/// `record()`, it is _unset_. Once `record()` has been called once,
/// it is _set_ and won't change anymore. This type is useful if you
/// need to track whether a feature was ever used during a session.
///
///
/// With `SerializationFormat::SimpleJson`, these histograms are
/// serialized as a plain number 0 (unset)/1 (set).
///
pub struct Flag {
    back_end: BackEnd<Plain>,

    /// A cache used to avoid spamming the Task once the flag has been set.
    cache: AtomicBool,
}

/// The storage, owned by the Telemetry Task.
struct FlagStorage {
    /// `true` once we have called `record`, `false` until then.
    encountered: bool
}

impl PlainRawStorage for FlagStorage {
    fn store(&mut self, _: u32) {
        self.encountered = true;
    }
    fn to_json(&self, format: &SerializationFormat) -> Json {
        match format {
            &SerializationFormat::SimpleJson => {
                Json::I64(if self.encountered { 1 } else { 0 })
            }
        }
    }
}

impl Histogram<()> for Flag {
    fn record_cb<F>(&self, cb: F) where F: FnOnce() -> Option<()>  {
        if self.cache.load(Ordering::Relaxed) {
            // Don't bother with dereferencing values or sending
            // messages, the histogram is already full.
            return;
        }
        if self.back_end.raw_record_cb(cb) {
            self.cache.store(true, Ordering::Relaxed);
        }
    }
}


impl Flag {
    ///
    /// Create a new Flag histogram with a given name.
    ///
    /// Argument `name` is used as key when processing and exporting
    /// the data. Each `name` must be unique to the `Service`.
    ///
    /// # Panics
    ///
    /// If `name` is already used by another histogram in `service`.
    ///
    pub fn new(service: &Service, name: String) -> Flag {
        let storage = Box::new(FlagStorage { encountered: false });
        let key = PrivateAccess::register_plain(service, name, storage);
        Flag {
            back_end: BackEnd::new(service, key),
            cache: AtomicBool::new(false),
        }
    }
}

impl Clone for Flag {
    fn clone(&self) -> Self {
        Flag {
            back_end: self.back_end.clone(),
            // The cache is not shared, but that's ok, it's just an
            // optimization.
            cache: AtomicBool::new(self.cache.load(Ordering::Relaxed)),
        }
    }
}

///
/// Linear histograms.
///
///
/// Linear histograms classify numeric integer values into same-sized
/// buckets. This type is typically used for percentages, or to store
/// a relatively precise approximation of the amount of resources
/// (time, memory) used by a section or a data structure.
///
///
/// With `SerializationFormat::SimpleJson`, these histograms are
/// serialized as an array of numbers, one per bucket, in the numeric
/// order of buckets.
pub struct Linear<T> where T: Flatten {
    witness: PhantomData<T>,
    back_end: BackEnd<Plain>,
}

impl<T> Histogram<T> for Linear<T> where T: Flatten {
    fn record_cb<F>(&self, cb: F) where F: FnOnce() -> Option<T>  {
        self.back_end.raw_record_cb(cb);
    }
}

impl<T> Linear<T> where T: Flatten {
    ///
    /// Create a new Linear histogram with a given name.
    ///
    /// - `name` is used as key when processing and exporting
    /// the data. Each `name` must be unique to the `Service`.
    ///
    /// - `min` is the minimal value expected to be entered in this
    /// histogram. Any value lower than `min` is rounded up to `min`.
    ///
    /// - `max` is the maximal value expected to be entered in this
    /// histogram. Any value higher than `max` is rounded up to `max`.
    ///
    /// - `buckets` is the number of buckets in this histogram. For
    /// highest possible precision, use `buckets = max - min + 1`.
    /// In most cases, however, such precision is not needed, so you
    /// should use a lower number of buckets.
    ///
    ///
    /// # Performance
    ///
    /// Increasing the number of buckets increases the memory usage on
    /// the client by a few bytes per bucket. More importantly, it also
    /// increases the size of the payload, hence the total amount of
    /// data that the application will eventually upload to a central
    /// server. If your application has many clients and you wish to
    /// keep your server happy and your bandwidth costs manageable,
    /// don't use too many buckets.
    ///
    ///
    /// # Panics
    ///
    /// If `name` is already used by another histogram in `service`.
    ///
    /// If `min >= max`.
    ///
    /// If `buckets < max - min + 1`.
    ///
    pub fn new(service: &Service, name: String, min: u32, max: u32, buckets: usize) -> Linear<T> {
        assert!(size_of::<u32>() <= size_of::<usize>());
        assert!(min < max);
        assert!(max - min + 1 >= buckets as u32);
        let shape = LinearBuckets::new(min, max, buckets);
        let storage = Box::new(LinearStorage::new(shape));
        let key = PrivateAccess::register_plain(service, name, storage);
        Linear {
            witness: PhantomData,
            back_end: BackEnd::new(service, key),
        }
    }
}

struct LinearStorage {
    values: Vec<u32>,// We cannot use an array here, as this would make the struct unsized.
    shape: LinearBuckets,
}


impl LinearStorage {
    fn new(shape: LinearBuckets) -> LinearStorage {
        let vec = vec_with_size(shape.buckets, 0);
        LinearStorage {
            values: vec,
            shape: shape,
        }
    }
}

impl PlainRawStorage for LinearStorage {
    fn store(&mut self, value: u32) {
        let index = self.shape.get_bucket(value);
        self.values[index] += 1;
    }
    fn to_json(&self, _: &SerializationFormat) -> Json {
        let json = Json::Array(self.values.iter().map(|&x| Json::I64(x as i64)).collect());
        json
    }
}

impl<T> Clone for Linear<T> where T: Flatten {
    fn clone(&self) -> Self {
        Linear {
            witness: PhantomData,
            back_end: self.back_end.clone()
        }
    }
}

///
///
/// Count histograms.
///
/// A Count histogram simply accumulates the numbers passed with
/// `record()`. Count histograms are useful, for instance, to know how
/// many times a feature has been used, or how many times an error has
/// been triggered.
///
///
/// With `SerializationFormat::SimpleJson`, these histograms are
/// serialized as a plain number.
///
#[derive(Clone)]
pub struct Count {
    back_end: BackEnd<Plain>,
}

// The storage, owned by the Telemetry Task.
struct CountStorage {
    value: u32
}

impl PlainRawStorage for CountStorage {
    fn store(&mut self, value: u32) {
        self.value += value;
    }
    fn to_json(&self, format: &SerializationFormat) -> Json {
        match format {
            &SerializationFormat::SimpleJson => {
                Json::I64(self.value as i64)
            }
        }
    }
}

impl Histogram<u32> for Count {
    fn record_cb<F>(&self, cb: F) where F: FnOnce() -> Option<u32>  {
        self.back_end.raw_record_cb(cb);
    }
}


impl Count {
    ///
    /// Create a new Count histogram with a given name.
    ///
    /// Argument `name` is used as key when processing and exporting
    /// the data. Each `name` must be unique to the `Service`.
    ///
    /// # Panics
    ///
    /// If `name` is already used by another histogram in `service`.
    ///
    pub fn new(service: &Service, name: String) -> Count {
        let storage = Box::new(CountStorage { value: 0 });
        let key = PrivateAccess::register_plain(service, name, storage);
        Count {
            back_end: BackEnd::new(service, key),
        }
    }
}



///
///
/// Enumerated histograms.
///
/// Enumerated histogram generalize Count histograms to families of
/// keys known at compile-time. They are useful, for instance, to know
/// how often users have picked a specific choice from several, or how
/// many times each kind of error has been triggered, etc.
///
///
/// With `SerializationFormat::SimpleJson`, these histograms are
/// serialized as an array of numbers, in the order of enum values.
///
pub struct Enum<K> where K: Flatten {
    witness: PhantomData<K>,
    back_end: BackEnd<Plain>,
}

// The storage, owned by the Telemetry Task.
struct EnumStorage {
    values: Vec<u32>
}

impl PlainRawStorage for EnumStorage {
    fn store(&mut self, value: u32) {
        vec_resize(&mut self.values, value as usize + 1, 0);
        self.values[value as usize] += 1;
    }
    fn to_json(&self, format: &SerializationFormat) -> Json {
        match format {
            &SerializationFormat::SimpleJson => {
                Json::Array(self.values.iter().map(|&x| Json::I64(x as i64)).collect())
            }
        }
    }
}

impl<K> Histogram<K> for Enum<K> where K: Flatten {
    fn record_cb<F>(&self, cb: F) where F: FnOnce() -> Option<K>  {
        self.back_end.raw_record_cb(cb);
    }
}


impl<K> Enum<K> where K: Flatten {
    ///
    /// Create a new Enum histogram with a given name.
    ///
    /// Argument `name` is used as key when processing and exporting
    /// the data. Each `name` must be unique to the `Service`.
    ///
    /// # Panics
    ///
    /// If `name` is already used by another histogram in `service`.
    ///
    pub fn new(service: &Service, name: String) -> Enum<K> {
        let storage = Box::new(EnumStorage { values: Vec::new() });
        let key = PrivateAccess::register_plain(service, name, storage);
        Enum {
            witness: PhantomData,
            back_end: BackEnd::new(service, key),
        }
    }
}

impl<K> Clone for Enum<K> where K: Flatten {
    fn clone(&self) -> Self {
        Enum {
            witness: PhantomData,
            back_end: self.back_end.clone()
        }
    }
}