-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathtodos.txt
executable file
·45 lines (25 loc) · 1.58 KB
/
todos.txt
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
CircleCI docs - configuration - https://circleci.com/docs/2.0/configuration-reference/
Use Docker with TravisCI: https://docs.travis-ci.com/user/docker/
* priority lock request should be put at front of queue
* provide option to just use lockUuid, and not unlock fn
* allow lock-requestors to force their way to obtaining a lock
similar to unlock-requestors being able to force unlock even if they don't have the key
* https://github.com/uWebSockets/uWebSockets
* unlock may be callable more than once, that should not happen.
* on disconnection we could delete wsClientId key/val from this.wsToKeys
but there should be no real need to do that since we won't have that many clients
/////////////////
created new benchmarks, you were right about some things - what really made a difference was using a random setTimeout,
using that, things started to favor Lockfile and Warlock again. However, there is an extremely non-linear relationship
between concurrency and performance, once you get concurrency above certain levels, the performance of lockfile and warlock
just plummets, whereas LM stays fairly linear. There's a tremendous difference there, night and day.
It's just a matter of how much concurrency you need. If you need a really high level of concurrency for your locking,
LM is way way better.
use:
"name": "@oresoftware/json-stream-parser",
"version": "0.0.1008",
do the F#
https://www.baeldung.com/java-nio2-async-socket-channel
if connection lost (server goes down)
client calls should have error passed to them, not just "end" event.
although the timeout might suffice.