this little bit of caching gives us a massive perf gain on Discourse

[SamSaffron]

Before:

sam@ubuntu:~/Source/active_model_serializers$ ab -c 1 -n 100 http://localhost:3000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking localhost (be patient).....done


Server Software:        thin
Server Hostname:        localhost
Server Port:            3000

Document Path:          /
Document Length:        44792 bytes

Concurrency Level:      1
Time taken for tests:   10.616 seconds
Complete requests:      100
Failed requests:        33
   (Connect: 0, Receive: 0, Length: 33, Exceptions: 0)
Write errors:           0
Total transferred:      4568084 bytes
HTML transferred:       4479163 bytes
Requests per second:    9.42 [#/sec] (mean)
Time per request:       106.160 [ms] (mean)
Time per request:       106.160 [ms] (mean, across all concurrent requests)
Transfer rate:          420.22 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0    0   0.0      0       0
Processing:    89  106  33.4     92     228
Waiting:       89  106  33.4     92     228
Total:         89  106  33.4     92     228

Percentage of the requests served within a certain time (ms)
  50%     92
  66%     96
  75%     97
  80%     98
  90%    156
  95%    198
  98%    204
  99%    228
 100%    228 (longest request)
sam@ubuntu:~/Source/active_model_serializers$ 

After:

sam@ubuntu:~/Source/active_model_serializers$ ab -c 1 -n 100 http://localhost:3000/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking localhost (be patient).....done


Server Software:        thin
Server Hostname:        localhost
Server Port:            3000

Document Path:          /
Document Length:        44792 bytes

Concurrency Level:      1
Time taken for tests:   8.599 seconds
Complete requests:      100
Failed requests:        18
   (Connect: 0, Receive: 0, Length: 18, Exceptions: 0)
Write errors:           0
Total transferred:      4568161 bytes
HTML transferred:       4479179 bytes
Requests per second:    11.63 [#/sec] (mean)
Time per request:       85.988 [ms] (mean)
Time per request:       85.988 [ms] (mean, across all concurrent requests)
Transfer rate:          518.80 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0    0   0.0      0       0
Processing:    75   86  28.1     77     188
Waiting:       75   86  28.1     77     188
Total:         75   86  28.1     77     188

Percentage of the requests served within a certain time (ms)
  50%     77
  66%     78
  75%     78
  80%     80
  90%     87
  95%    179
  98%    182
  99%    188
 100%    188 (longest request)

---

So in English, median request has gone down for 106ms to 85ms. A huge gain.

This happens cause internally, AM calls pluralize over and over again ... and pluralize is slow

https://github.com/rails-api/active_model_serializers/blob/master/lib/active_model/serializer/associations.rb#L157

As isolated by the yet to be announced MiniProfiler flame graph:

[ZogStriP]

👍

[steveklabnik]

✂️

[SamSaffron]

sorry about that ... will sort it out, took a lot of dicking to find a clean patch

[steveklabnik]

No worries. :D

[steveklabnik]

Seems good. :)

[SamSaffron]

ok 🍦 we should be good now ... also has anyone looked at: #241 I really want to get us on latest am serializer, not sure what the best approach is to fix it

[steveklabnik]

I have AMS work on my plate for tomorrow, so I haven't looked into it yet.

[steveklabnik]

❤️