talks/scala-channels.html


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

<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml" lang="en-US" xml:lang="en-US">
<head>
  <meta charset="utf-8" />
  <meta name="generator" content="pandoc" />
  <meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=yes" />
  <meta name="author" content="Jakob Odersky" />
  <meta name="dcterms.date" content="2018-11-15" />
  <title>Channeling the Inner Complexity</title>
  <style type="text/css">
      code{white-space: pre-wrap;}
      span.smallcaps{font-variant: small-caps;}
      span.underline{text-decoration: underline;}
      div.column{display: inline-block; vertical-align: top; width: 50%;}
  </style>
  <style type="text/css">
a.sourceLine { display: inline-block; line-height: 1.25; }
a.sourceLine { pointer-events: none; color: inherit; text-decoration: inherit; }
a.sourceLine:empty { height: 1.2em; position: absolute; }
.sourceCode { overflow: visible; }
code.sourceCode { white-space: pre; position: relative; }
div.sourceCode { margin: 1em 0; }
pre.sourceCode { margin: 0; }
@media screen {
div.sourceCode { overflow: auto; }
}
@media print {
code.sourceCode { white-space: pre-wrap; }
a.sourceLine { text-indent: -1em; padding-left: 1em; }
}
pre.numberSource a.sourceLine
  { position: relative; }
pre.numberSource a.sourceLine:empty
  { position: absolute; }
pre.numberSource a.sourceLine::before
  { content: attr(data-line-number);
    position: absolute; left: -5em; text-align: right; vertical-align: baseline;
    border: none; pointer-events: all;
    -webkit-touch-callout: none; -webkit-user-select: none;
    -khtml-user-select: none; -moz-user-select: none;
    -ms-user-select: none; user-select: none;
    padding: 0 4px; width: 4em;
  }
pre.numberSource { margin-left: 3em;  padding-left: 4px; }
div.sourceCode
  { color: #cccccc; background-color: #303030; }
@media screen {
a.sourceLine::before { text-decoration: underline; }
}
code span.al { color: #ffcfaf; } /* Alert */
code span.an { color: #7f9f7f; font-weight: bold; } /* Annotation */
code span.at { } /* Attribute */
code span.bn { color: #dca3a3; } /* BaseN */
code span.bu { } /* BuiltIn */
code span.cf { color: #f0dfaf; } /* ControlFlow */
code span.ch { color: #dca3a3; } /* Char */
code span.cn { color: #dca3a3; font-weight: bold; } /* Constant */
code span.co { color: #7f9f7f; } /* Comment */
code span.cv { color: #7f9f7f; font-weight: bold; } /* CommentVar */
code span.do { color: #7f9f7f; } /* Documentation */
code span.dt { color: #dfdfbf; } /* DataType */
code span.dv { color: #dcdccc; } /* DecVal */
code span.er { color: #c3bf9f; } /* Error */
code span.ex { } /* Extension */
code span.fl { color: #c0bed1; } /* Float */
code span.fu { color: #efef8f; } /* Function */
code span.im { } /* Import */
code span.in { color: #7f9f7f; font-weight: bold; } /* Information */
code span.kw { color: #f0dfaf; } /* Keyword */
code span.op { color: #f0efd0; } /* Operator */
code span.ot { color: #efef8f; } /* Other */
code span.pp { color: #ffcfaf; font-weight: bold; } /* Preprocessor */
code span.sc { color: #dca3a3; } /* SpecialChar */
code span.ss { color: #cc9393; } /* SpecialString */
code span.st { color: #cc9393; } /* String */
code span.va { } /* Variable */
code span.vs { color: #cc9393; } /* VerbatimString */
code span.wa { color: #7f9f7f; font-weight: bold; } /* Warning */
  </style>
</head>
<body>
<header>
<h1 class="title">Channeling the Inner Complexity</h1>
<p class="subtitle">or, lightweight threads and channels for Scala</p>
<p class="author">Jakob Odersky</p>
<p class="date">2018-11-15</p>
</header>
<h1 id="overview">Overview</h1>
<ul>
<li>Basic concurrency models</li>
<li>Futures and Promises</li>
<li>Channels and lightweight threads</li>
</ul>
<h1 id="definitions">Definitions</h1>
<ul>
<li><p><strong>parallelism</strong>: the simultaneous execution on multiple processors of different parts of a program<a href="#fn1" class="footnote-ref" id="fnref1"><sup>1</sup></a></p></li>
<li><p><strong>concurrency</strong>: the ability of different parts of a program to be executed out-of-order or in partial order, without affecting the final outcome<a href="#fn2" class="footnote-ref" id="fnref2"><sup>2</sup></a></p></li>
</ul>
<h1 id="premise">Premise</h1>
<ul>
<li><p><strong>scalable programs need a good concurrency model</strong></p></li>
<li>“good”:
<ul>
<li>increased efficiency (take advantage of parallelism)</li>
<li>reduced complexity</li>
</ul></li>
</ul>
<hr />
<h1 id="concurrency---threads">Concurrency - Threads</h1>
<ul>
<li><p>single entry point, sequence of instructions</p></li>
<li><p>traditional way to decompose programs for parallel execution</p></li>
<li><p>own stack and kernel resources (fairly expensive)</p></li>
<li><p>context switches (fairly expensive)</p></li>
<li><p>runnable on a physical processor</p></li>
</ul>
<hr />
<h1 id="single-thread">Single Thread</h1>
<div class="sourceCode" id="cb1"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb1-1" data-line-number="1"><span class="kw">def</span> <span class="fu">mkmeme</span>(imageUrl: String, text: String): Image = {</a>
<a class="sourceLine" id="cb1-2" data-line-number="2">  <span class="kw">val</span> layer1: Image = <span class="fu">fetchUrl</span>(imageUrl) <span class="co">// network call</span></a>
<a class="sourceLine" id="cb1-3" data-line-number="3">  <span class="kw">val</span> layer2: Image = <span class="fu">textToImage</span>(text) <span class="co">// slow</span></a>
<a class="sourceLine" id="cb1-4" data-line-number="4">  <span class="fu">superimpose</span>(layer1, layer2) <span class="co">// need both results</span></a>
<a class="sourceLine" id="cb1-5" data-line-number="5">}</a></code></pre></div>
<hr />
<h1 id="single-thread-1">Single Thread</h1>
<ul>
<li>concurrency unit is the whole program</li>
</ul>
<hr />
<h1 id="many-threads">Many Threads</h1>
<div class="sourceCode" id="cb2"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb2-1" data-line-number="1"><span class="kw">def</span> <span class="fu">mkmeme</span>(imageUrl: String, text: String): Image = {</a>
<a class="sourceLine" id="cb2-2" data-line-number="2">  <span class="kw">var</span> layer1: Image = <span class="kw">null</span></a>
<a class="sourceLine" id="cb2-3" data-line-number="3">  <span class="kw">var</span> layer2: Image = <span class="kw">null</span></a>
<a class="sourceLine" id="cb2-4" data-line-number="4">  thread {</a>
<a class="sourceLine" id="cb2-5" data-line-number="5">    layer1 = <span class="fu">fetchUrl</span>(imageUrl)</a>
<a class="sourceLine" id="cb2-6" data-line-number="6">  }</a>
<a class="sourceLine" id="cb2-7" data-line-number="7">  thread {</a>
<a class="sourceLine" id="cb2-8" data-line-number="8">    layer2 = <span class="fu">textToImage</span>(text)</a>
<a class="sourceLine" id="cb2-9" data-line-number="9">  }</a>
<a class="sourceLine" id="cb2-10" data-line-number="10">  <span class="kw">while</span>(layer1 == <span class="kw">null</span> || layer2 == <span class="kw">null</span>) {</a>
<a class="sourceLine" id="cb2-11" data-line-number="11">    <span class="co">// wait somehow</span></a>
<a class="sourceLine" id="cb2-12" data-line-number="12">  }</a>
<a class="sourceLine" id="cb2-13" data-line-number="13">  <span class="fu">superimpose</span>(layer1, layer2)</a>
<a class="sourceLine" id="cb2-14" data-line-number="14">}</a></code></pre></div>
<hr />
<h1 id="many-threads-1">Many Threads</h1>
<ul>
<li><p>synchronization between threads at some point</p></li>
<li><p>rendezvous through memory barriers (CMPXCHG)</p></li>
<li><p>logic flow much more complex</p></li>
<li><p>threads, blocked and running</p>
<ul>
<li>consume memory</li>
<li>memory is cheap! create more threads? context switches</li>
</ul></li>
<li><p>threads are a low-level building block, using them efficiently is complex</p></li>
<li><p>not available on all platforms (i.e. browser)</p></li>
</ul>
<hr />
<h1 id="multiple-threads-queue-based">Multiple Threads, Queue-based</h1>
<div class="sourceCode" id="cb3"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb3-1" data-line-number="1"><span class="kw">def</span> <span class="fu">mkmeme</span>(imageUrl: String, text: String): Image = {</a>
<a class="sourceLine" id="cb3-2" data-line-number="2">  <span class="kw">val</span> q1 = Queue[Image]</a>
<a class="sourceLine" id="cb3-3" data-line-number="3">  <span class="kw">val</span> q2 = Queue[Image]</a>
<a class="sourceLine" id="cb3-4" data-line-number="4">  thread {</a>
<a class="sourceLine" id="cb3-5" data-line-number="5">    q1.<span class="fu">put</span>(<span class="fu">fetchUrl</span>(imageUrl))</a>
<a class="sourceLine" id="cb3-6" data-line-number="6">  }</a>
<a class="sourceLine" id="cb3-7" data-line-number="7">  thread {</a>
<a class="sourceLine" id="cb3-8" data-line-number="8">    q2.<span class="fu">put</span>(<span class="fu">textToImage</span>(text))</a>
<a class="sourceLine" id="cb3-9" data-line-number="9">  }</a>
<a class="sourceLine" id="cb3-10" data-line-number="10">  <span class="fu">superimpose</span>(q1.<span class="fu">take</span>(), q2.<span class="fu">take</span>())</a>
<a class="sourceLine" id="cb3-11" data-line-number="11">}</a></code></pre></div>
<h1 id="multiple-threads-queue-based-1">Multiple Threads, Queue-based</h1>
<ul>
<li>simpler logic flow</li>
<li>same resource usage as plain threads</li>
</ul>
<hr />
<h1 id="concurrency---callbacks">Concurrency - Callbacks</h1>
<ul>
<li><p>“reactive”</p></li>
<li><p>many entrypoints</p></li>
<li><p>register operation on event</p></li>
<li><p>“call back” when event has happened, operation is run</p></li>
<li>examples:
<ul>
<li>JavaScript</li>
<li>libuv</li>
<li>event loops</li>
</ul></li>
<li><p>in a sense, a more fundamental construct</p></li>
</ul>
<p>–</p>
<h1 id="callbacks">Callbacks</h1>
<div class="sourceCode" id="cb4"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb4-1" data-line-number="1"><span class="kw">def</span> <span class="fu">mkmeme</span>(imageUrl: String, text: String,</a>
<a class="sourceLine" id="cb4-2" data-line-number="2">    callback: Image =&gt; Unit): Unit = {</a>
<a class="sourceLine" id="cb4-3" data-line-number="3">  <span class="kw">var</span> layer1 = <span class="kw">null</span></a>
<a class="sourceLine" id="cb4-4" data-line-number="4">  <span class="kw">var</span> layer2 = <span class="kw">null</span></a>
<a class="sourceLine" id="cb4-5" data-line-number="5">  <span class="kw">def</span> <span class="fu">combine</span>() = <span class="fu">callback</span>(<span class="fu">superimpose</span>(layer1, layer2))</a>
<a class="sourceLine" id="cb4-6" data-line-number="6">  <span class="fu">fetchUrl</span>(imageUrl, img =&gt; {</a>
<a class="sourceLine" id="cb4-7" data-line-number="7">    layer1 = img</a>
<a class="sourceLine" id="cb4-8" data-line-number="8">    <span class="kw">if</span> (layer2 != <span class="kw">null</span>) { <span class="co">//!\\ danger if parallelism &gt; 1</span></a>
<a class="sourceLine" id="cb4-9" data-line-number="9">      <span class="fu">combine</span>()</a>
<a class="sourceLine" id="cb4-10" data-line-number="10">    }</a>
<a class="sourceLine" id="cb4-11" data-line-number="11">  })</a>
<a class="sourceLine" id="cb4-12" data-line-number="12">  <span class="fu">textToImage</span>(text, img =&gt; {</a>
<a class="sourceLine" id="cb4-13" data-line-number="13">    layer2 = img</a>
<a class="sourceLine" id="cb4-14" data-line-number="14">    <span class="kw">if</span> (layer1 != <span class="kw">null</span>) {</a>
<a class="sourceLine" id="cb4-15" data-line-number="15">      <span class="fu">combine</span>()</a>
<a class="sourceLine" id="cb4-16" data-line-number="16">    }</a>
<a class="sourceLine" id="cb4-17" data-line-number="17">  })</a>
<a class="sourceLine" id="cb4-18" data-line-number="18">}</a></code></pre></div>
<hr />
<p><img src="data:image/jpeg;base64,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" alt="callback" /><br />
</p>
<hr />
<h1 id="callbacks-1">Callbacks</h1>
<ul>
<li><p>advantages:</p>
<ul>
<li>little resource overhead</li>
<li>available on all platforms</li>
<li>runnable on many processors</li>
</ul></li>
<li>disadvantage:
<ul>
<li>program logic quickly becomes extremely complex and scattered: <em>callback hell</em></li>
</ul></li>
</ul>
<hr />
<ul>
<li><p>can we wrap callbacks in a more functional way?</p>
<ul>
<li>reduce complexity</li>
<li>keep efficiency, and run it on ideal number of processors</li>
</ul></li>
</ul>
<hr />
<h1 id="concurrency---futures">Concurrency - Futures</h1>
<h2 id="scala.concurrent.futurea"><code>scala.concurrent.Future[A]</code></h2>
<ul>
<li><p>contains an operation of result type <code>A</code></p></li>
<li><p>transformable with <code>map</code> and <code>flatMap</code></p></li>
<li><p>when operation is run, future completes with a result (success or failure)</p></li>
</ul>
<hr />
<h1 id="future">Future</h1>
<div class="sourceCode" id="cb5"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb5-1" data-line-number="1"><span class="kw">def</span> <span class="fu">mkmeme</span>(imageUrl: String, text: String): Future[Image] = {</a>
<a class="sourceLine" id="cb5-2" data-line-number="2">  <span class="kw">val</span> layer1: Future[Image] = <span class="fu">fetchUrl</span>(imageUrl) </a>
<a class="sourceLine" id="cb5-3" data-line-number="3">  <span class="kw">val</span> layer2: Future[Image] = <span class="fu">textToImage</span>(text)</a>
<a class="sourceLine" id="cb5-4" data-line-number="4">  <span class="kw">for</span> {</a>
<a class="sourceLine" id="cb5-5" data-line-number="5">    l1 &lt;- layer1</a>
<a class="sourceLine" id="cb5-6" data-line-number="6">    l2 &lt;- layer2</a>
<a class="sourceLine" id="cb5-7" data-line-number="7">  } <span class="kw">yield</span> {</a>
<a class="sourceLine" id="cb5-8" data-line-number="8">    <span class="fu">superimpose</span>(l1, l2)</a>
<a class="sourceLine" id="cb5-9" data-line-number="9">  }</a>
<a class="sourceLine" id="cb5-10" data-line-number="10">}</a></code></pre></div>
<hr />
<h1 id="promises">Promises</h1>
<h2 id="scala.concurren.promisea"><code>scala.concurren.Promise[A]</code></h2>
<ul>
<li>used to create and complete futures at the edge of the callback graph</li>
</ul>
<hr />
<div class="sourceCode" id="cb6"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb6-1" data-line-number="1"><span class="co">// ScalaJS, env: browser</span></a>
<a class="sourceLine" id="cb6-2" data-line-number="2"></a>
<a class="sourceLine" id="cb6-3" data-line-number="3"><span class="kw">def</span> url: Future[String] = {</a>
<a class="sourceLine" id="cb6-4" data-line-number="4">  <span class="kw">val</span> promise = Promise[String] <span class="co">// create promise</span></a>
<a class="sourceLine" id="cb6-5" data-line-number="5">  input.<span class="fu">onsubmit</span>(_ =&gt; promise.<span class="fu">success</span>(input.<span class="fu">value</span>))</a>
<a class="sourceLine" id="cb6-6" data-line-number="6">  promise.<span class="fu">future</span></a>
<a class="sourceLine" id="cb6-7" data-line-number="7">}</a>
<a class="sourceLine" id="cb6-8" data-line-number="8"></a>
<a class="sourceLine" id="cb6-9" data-line-number="9"><span class="co">// single callback at the edge</span></a>
<a class="sourceLine" id="cb6-10" data-line-number="10">url.<span class="fu">map</span>(fetch).<span class="fu">onComplete</span>{</a>
<a class="sourceLine" id="cb6-11" data-line-number="11">  <span class="kw">case</span> <span class="fu">Success</span>(site) =&gt; webview.<span class="fu">value</span> = site</a>
<a class="sourceLine" id="cb6-12" data-line-number="12">  <span class="kw">case</span> <span class="fu">Failure</span>(error) =&gt;</a>
<a class="sourceLine" id="cb6-13" data-line-number="13">    textbox.<span class="fu">value</span> = <span class="st">&quot;oh no!&quot;</span></a>
<a class="sourceLine" id="cb6-14" data-line-number="14">    textbox.<span class="fu">color</span> = red</a>
<a class="sourceLine" id="cb6-15" data-line-number="15">}</a></code></pre></div>
<h1 id="execution-contexts">Execution Contexts</h1>
<p>Who runs a future?</p>
<ul>
<li>one process traverses all callbacks? no!</li>
<li>operation “chunks” on an execution context</li>
</ul>
<h2 id="executioncontext"><code>ExecutionContext</code></h2>
<ul>
<li><p>contains graph of callbacks as chunks</p>
<div class="sourceCode" id="cb7"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb7-1" data-line-number="1">future1.<span class="fu">flatMap</span>(f1 =&gt; <span class="fu">op1</span>(f1).<span class="fu">map</span>(<span class="fu">op2</span>(_))(ec))(ec)</a></code></pre></div></li>
<li><p>chunks are run on a <em>ThreadPool</em></p></li>
</ul>
<h2 id="threadpool"><code>ThreadPool</code></h2>
<ul>
<li><p>(limited) group of threads</p></li>
<li><p>every thread runs a chunk, when done takes a next chunk</p>
<ul>
<li>aside: <em>when done</em> <span class="math inline">←</span> this is why blocking in futures is not recomended</li>
</ul></li>
</ul>
<hr />
<h1 id="futures---composition">Futures - Composition</h1>
<div class="sourceCode" id="cb8"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb8-1" data-line-number="1"><span class="kw">def</span> <span class="fu">lookupUser</span>(id: String): Future[Option[User]]</a>
<a class="sourceLine" id="cb8-2" data-line-number="2"><span class="kw">def</span> <span class="fu">authorize</span>(user: User, capabilities: Set[Cap]):</a>
<a class="sourceLine" id="cb8-3" data-line-number="3">  Future[Option[User]]</a>
<a class="sourceLine" id="cb8-4" data-line-number="4"></a>
<a class="sourceLine" id="cb8-5" data-line-number="5"><span class="kw">def</span> <span class="fu">authorizeduser</span>(userId: String): Future[Option[User]] = {</a>
<a class="sourceLine" id="cb8-6" data-line-number="6">  <span class="fu">lookupUser</span>(userId).<span class="fu">flatMap</span>{</a>
<a class="sourceLine" id="cb8-7" data-line-number="7">    <span class="kw">case</span> None =&gt; Future.<span class="fu">successful</span>(None)</a>
<a class="sourceLine" id="cb8-8" data-line-number="8">    <span class="kw">case</span> Some(user) =&gt; <span class="fu">authorize</span>(user, Set(<span class="st">&quot;see_meme&quot;</span>))</a>
<a class="sourceLine" id="cb8-9" data-line-number="9">  }</a>
<a class="sourceLine" id="cb8-10" data-line-number="10">}</a></code></pre></div>
<hr />
<h1 id="futures---shortcomings">Futures - Shortcomings</h1>
<ol type="1">
<li><p>composition can be messy<a href="#fn3" class="footnote-ref" id="fnref3"><sup>3</sup></a></p></li>
<li><p>one-shot; it is not simple to model recurrent events</p></li>
</ol>
<hr />
<h1 id="solution-to-1---scala-async">Solution to 1 - Scala Async</h1>
<ul>
<li><p>Can we write a program that looks synchronous (single-threaded), but is split into chunks and run on a thread pool?</p></li>
<li><p>yes, with macros!</p></li>
<li><p>two constructs:</p>
<ul>
<li><code>async(a: =&gt; A): Future[A]</code> // macro</li>
<li><code>await(f: Future[A]): A</code> // usable only in await</li>
</ul></li>
<li><p>installs handlers on futures to run a state machine</p></li>
<li><p>official project of the Scala Center</p></li>
<li><p><a href="https://github.com/scala/scala-async" class="uri">https://github.com/scala/scala-async</a></p></li>
<li><p>see also python async</p></li>
</ul>
<hr />
<div class="sourceCode" id="cb9"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb9-1" data-line-number="1"><span class="kw">import</span> scala.<span class="fu">concurrent</span>.<span class="fu">ExecutionContext</span>.<span class="fu">Implicits</span>.<span class="fu">global</span></a>
<a class="sourceLine" id="cb9-2" data-line-number="2"><span class="kw">import</span> scala.<span class="fu">async</span>.<span class="fu">Async</span>._</a>
<a class="sourceLine" id="cb9-3" data-line-number="3"></a>
<a class="sourceLine" id="cb9-4" data-line-number="4"><span class="co">// looks like single-threaded code</span></a>
<a class="sourceLine" id="cb9-5" data-line-number="5"><span class="kw">def</span> <span class="fu">mkmeme</span>(imageUrl: String, text: String): Future[Image] = </a>
<a class="sourceLine" id="cb9-6" data-line-number="6">  async {</a>
<a class="sourceLine" id="cb9-7" data-line-number="7">    <span class="kw">val</span> layer1 = <span class="fu">await</span>(<span class="fu">fetchUrl</span>(imageUrl))</a>
<a class="sourceLine" id="cb9-8" data-line-number="8">    <span class="kw">val</span> layer2 = <span class="fu">await</span>(<span class="fu">textToImage</span>(text))</a>
<a class="sourceLine" id="cb9-9" data-line-number="9">    <span class="fu">superimpose</span>(layer1, layer2)</a>
<a class="sourceLine" id="cb9-10" data-line-number="10">  }</a></code></pre></div>
<hr />
<h1 id="solution-to-2---channels">Solution to 2 - Channels</h1>
<ul>
<li><p>futures are one-shot value</p></li>
<li><p>queues are general useful construct for scalable programs</p>
<ul>
<li>separation of concerns</li>
</ul></li>
<li><p>as shown previously, traditional thread-based queues block</p></li>
<li><p>can we avoid blocking, yet keep the programming model?</p></li>
</ul>
<h1 id="solution-to-2---channels-1">Solution to 2 - Channels</h1>
<ul>
<li><p>project “escale” (fr. stop, as in bus stop)</p></li>
<li><p>inspired from Clojure’s core.async library</p></li>
<li><p>watch Rich Hickey’s talk about it <a href="https://www.infoq.com/presentations/core-async-clojure" class="uri">https://www.infoq.com/presentations/core-async-clojure</a></p></li>
<li>constructs:
<ul>
<li><code>go {block}: Future[A]</code> ~ lightweight thread</li>
<li><code>Channel[A]</code> ~ queue</li>
<li><code>ch.put(value: A): Future[A]</code> ~ write operation</li>
<li><code>ch.take(): Future[A]</code> ~ read operation</li>
<li><code>select(ch: Channel[_]*)</code></li>
</ul></li>
<li><p>syntax sugar</p></li>
<li><p>form of communicating sequential processes (CSP) <span class="citation" data-cites="csp">[1]</span></p>
<ul>
<li>there is a formal mathematical model</li>
</ul></li>
<li><p>since runtime is abstracted, runs on JVM, JS and Native</p></li>
</ul>
<hr />
<h1 id="escale">escale</h1>
<div class="sourceCode" id="cb10"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb10-1" data-line-number="1"><span class="kw">import</span> scala.<span class="fu">concurrent</span>.<span class="fu">ExecutionContext</span>.<span class="fu">Implicits</span>.<span class="fu">global</span></a>
<a class="sourceLine" id="cb10-2" data-line-number="2"><span class="kw">import</span> escale.<span class="fu">syntax</span>._</a>
<a class="sourceLine" id="cb10-3" data-line-number="3"></a>
<a class="sourceLine" id="cb10-4" data-line-number="4"><span class="kw">val</span> ch = chan[Int]() <span class="co">// create a channel</span></a>
<a class="sourceLine" id="cb10-5" data-line-number="5"></a>
<a class="sourceLine" id="cb10-6" data-line-number="6">go {</a>
<a class="sourceLine" id="cb10-7" data-line-number="7">  ch !&lt; <span class="dv">1</span> <span class="co">// write to channel, &quot;block&quot; if no room</span></a>
<a class="sourceLine" id="cb10-8" data-line-number="8">  <span class="fu">println</span>(<span class="st">&quot;wrote 1&quot;</span>)</a>
<a class="sourceLine" id="cb10-9" data-line-number="9">}</a>
<a class="sourceLine" id="cb10-10" data-line-number="10">go {</a>
<a class="sourceLine" id="cb10-11" data-line-number="11">  ch !&lt; <span class="dv">2</span></a>
<a class="sourceLine" id="cb10-12" data-line-number="12">  <span class="fu">println</span>(<span class="st">&quot;wrote 2&quot;</span>)</a>
<a class="sourceLine" id="cb10-13" data-line-number="13">}</a>
<a class="sourceLine" id="cb10-14" data-line-number="14"></a>
<a class="sourceLine" id="cb10-15" data-line-number="15">go {</a>
<a class="sourceLine" id="cb10-16" data-line-number="16">  <span class="kw">val</span> r: Int = !&lt;(ch) <span class="co">// read from channel</span></a>
<a class="sourceLine" id="cb10-17" data-line-number="17">  <span class="fu">println</span>(r)</a>
<a class="sourceLine" id="cb10-18" data-line-number="18">  <span class="fu">println</span>(!&lt;(ch))</a>
<a class="sourceLine" id="cb10-19" data-line-number="19">}</a></code></pre></div>
<hr />
<h1 id="escale-1">escale</h1>
<div class="sourceCode" id="cb11"><pre class="sourceCode scala"><code class="sourceCode scala"><a class="sourceLine" id="cb11-1" data-line-number="1"><span class="kw">import</span> escale.<span class="fu">syntax</span>._</a>
<a class="sourceLine" id="cb11-2" data-line-number="2"></a>
<a class="sourceLine" id="cb11-3" data-line-number="3">go {</a>
<a class="sourceLine" id="cb11-4" data-line-number="4">  <span class="kw">val</span> Ch1 = chan[Int]() <span class="co">// create a channel</span></a>
<a class="sourceLine" id="cb11-5" data-line-number="5">  <span class="kw">val</span> Ch2 = chan[Int]()</a>
<a class="sourceLine" id="cb11-6" data-line-number="6">  </a>
<a class="sourceLine" id="cb11-7" data-line-number="7">  go { Ch1 !&lt; <span class="dv">1</span> } <span class="co">// write to channel</span></a>
<a class="sourceLine" id="cb11-8" data-line-number="8">  go { Ch2 !&lt; <span class="dv">1</span> }</a>
<a class="sourceLine" id="cb11-9" data-line-number="9"></a>
<a class="sourceLine" id="cb11-10" data-line-number="10">  <span class="co">// &quot;await&quot; one and only one value</span></a>
<a class="sourceLine" id="cb11-11" data-line-number="11">  <span class="fu">select</span>(Ch1, Ch2) <span class="kw">match</span> {</a>
<a class="sourceLine" id="cb11-12" data-line-number="12">    <span class="kw">case</span> (Ch1, value) =&gt; <span class="st">&quot;ch1 was first&quot;</span></a>
<a class="sourceLine" id="cb11-13" data-line-number="13">    <span class="kw">case</span> (Ch2, value) =&gt; <span class="st">&quot;ch2 was first&quot;</span></a>
<a class="sourceLine" id="cb11-14" data-line-number="14">  }</a>
<a class="sourceLine" id="cb11-15" data-line-number="15">}</a></code></pre></div>
<h1 id="escale---implementation">escale - Implementation</h1>
<ul>
<li><p>proof-of-concept</p></li>
<li><p>https://github.com/jodersky/escale (soon)</p></li>
<li><p>channels take care of buffering and efficient locking operations</p></li>
<li><p>put and take return futures (select slightly more complex, but also returns a future)</p></li>
<li><p>rely on scala-async to transform future into state machine</p></li>
<li><p>provide syntax sugar to hide calls to <code>await</code> and alias <code>async</code></p></li>
</ul>
<h1 id="escale---roadmap">escale - Roadmap</h1>
<ul>
<li><p>channel closing and error handling</p></li>
<li><p>deeper integration with scala async</p>
<ul>
<li>explore working with the state machine directly, rather than relying on double macro transformations</li>
</ul></li>
<li><p>select on puts</p></li>
<li><p>buffer policies (drop first, sliding window)</p></li>
<li><p>API improvements:</p>
<ul>
<li>consider replacing symbols</li>
<li>remove wilcard <code>import escale.sytntax._</code></li>
<li>directionality type refinements</li>
</ul></li>
</ul>
<hr />
<h1 id="summary-what-have-we-done">Summary: what have we done?</h1>
<ul>
<li>replaced queues and threads with conceptually lightweight queues and threads</li>
<li>same programming model, better concurrency</li>
<li>in a library!</li>
</ul>
<p><em>All problems in computer science can be solved by another level of indirection</em>.</p>
<hr />
<h1 id="other-approaches">Other Approaches</h1>
<h2 id="actors">Actors</h2>
<ul>
<li><p>actors and CSP can be considered duals</p></li>
<li><p>actors are named, processes are anonymous</p></li>
<li><p>message path is anonymous, channels are named</p></li>
<li><p>sending messages is fundamentally non-blocking, whereas (unbuffered) channels can serve as rendezvous points</p></li>
</ul>
<h2 id="reactive-streams">Reactive Streams</h2>
<ul>
<li>builds a protocol on top of actors to achieve rendezvous capabilities and backpressure</li>
</ul>
<hr />
<h1 id="guidelines">Guidelines</h1>
<p>Keep programs <em>simple</em>, it will make it <em>easier</em> for others to understand.</p>
<ol type="1">
<li>write synchronous logic</li>
<li>use futures and promises with scala-async</li>
<li>escale and other concurrency libraries</li>
<li>…</li>
<li>…</li>
<li>…</li>
<li>…</li>
<li>…</li>
<li>…</li>
<li>consider callbacks</li>
</ol>
<hr />
<h1 id="thank-you">Thank You!</h1>
<ul>
<li><p>slides: https://jakob.odersky.com/talks</p></li>
<li><p>project: https://github.com/jodersky/escale</p></li>
<li><p>author: <code>@jodersky</code></p></li>
</ul>
<hr />
<h1 id="references" class="unnumbered">References</h1>
<div id="refs" class="references">
<div id="ref-csp">
<p>[1] C. A. R. Hoare, “Communicating sequential processes,” <em>Communications of the ACM. 21 (8)</em>, pp. 666–667, 1978.</p>
</div>
</div>
<section class="footnotes">
<hr />
<ol>
<li id="fn1"><p>https://en.wikipedia.org/wiki/Parallelism<a href="#fnref1" class="footnote-back">↩</a></p></li>
<li id="fn2"><p>https://en.wikipedia.org/wiki/Concurrency_(computer_science)<a href="#fnref2" class="footnote-back">↩</a></p></li>
<li id="fn3"><p>monad transformers may help<a href="#fnref3" class="footnote-back">↩</a></p></li>
</ol>
</section>
</body>
</html>