support json payload template #202
Conversation
qzhuyan
force-pushed
the
dev/william/json-payload-template
branch
from
dfe1c97 to
3861b79
Compare
qzhuyan
force-pushed
the
dev/william/json-payload-template
branch
from
13c30a9 to
7ab26de
Compare
qzhuyan
force-pushed
the
dev/william/json-payload-template
branch
from
7ab26de to
f2e942e
Compare
| @@ -25,6 +25,7 @@ Profiles = {profiles,[ {escript, []} | |||
| , getopt | |||
| , gun | |||
| , cowlib | |||
| , jsx | |||
There was a problem hiding this comment.
switch to 27 and use https://www.erlang.org/doc/apps/stdlib/json.html?
There was a problem hiding this comment.
true.
json encoding with large payload is toooo slow. Took me some time to find the slowness is coming from jsx.
In my demo, or test report, I will use eterm encoding, json encoding is reserved for specific customer.
topic_spec.json
Outdated
| "QoS": 1, | ||
| "payload_encoding": "eterm", | ||
| "payload": {"foo1" : "bar1", "timestamp": "0", "Data": "10"}, | ||
| "stream" : 0, |
There was a problem hiding this comment.
stream 0 is default, signal stream solution used by TCP TCP/TLS and QUIC single stream.
stream >0 is for multi stream features and will only work with QUIC enabled .
| "interval_ms": "5000", | ||
| "inject_timestamp" : "ms", | ||
| "QoS": 1, | ||
| "payload_encoding": "eterm", |
There was a problem hiding this comment.
could be json but slow for large payload.
qzhuyan
force-pushed
the
dev/william/json-payload-template
branch
from
ff56212 to
6e1000a
Compare
rebar.config.script
Outdated
| [emqtt_bench, | ||
| prometheus, | ||
| quantile_estimator, | ||
| cowboy |
There was a problem hiding this comment.
not necessary, I added it in emqtt_bench.app.src
There was a problem hiding this comment.
but isn't this to generate the escript, which will need jsx to decode the template files?
There was a problem hiding this comment.
It is working as expected at: https://github.com/qzhuyan/emqtt-bench/releases/tag/0.4.22%2Bcustm1
I think rebar pack the jsx into the escript archive file if it is defined in the app.src.
maybe we should put these (emqtt_bench, prometheus, quantile_estimator, cowboy) to app src as well?
I think escript_incl_apps is to include none dependable apps such like 'redbug'
There was a problem hiding this comment.
It is working as expected at: https://github.com/qzhuyan/emqtt-bench/releases/tag/0.4.22%2Bcustm1
I think rebar pack the jsx into the escript archive file if it is defined in app.src.
maybe we should put these (emqtt_bench, prometheus, quantile_estimator, cowboy) to app src as well?
There was a problem hiding this comment.
I see, I thought this setting could be relevant.
maybe we should put these (emqtt_bench, prometheus, quantile_estimator, cowboy) to app src as well?
I think they are (except the first, which is the app itself). Though quicer would need to be conditionally inserted as well?
rebar.config.script
Outdated
| [emqtt_bench, | ||
| prometheus, | ||
| quantile_estimator, | ||
| cowboy |
There was a problem hiding this comment.
I see, I thought this setting could be relevant.
maybe we should put these (emqtt_bench, prometheus, quantile_estimator, cowboy) to app src as well?
I think they are (except the first, which is the app itself). Though quicer would need to be conditionally inserted as well?
qzhuyan
force-pushed
the
dev/william/json-payload-template
branch
from
072a53d to
fb45cab
Compare
Not sure if we should merge this.this branch used a private fork of EMQTT and intro some features that might not that be useful for others.This branch added new arg
--topics_payload some.json, the json file has defined topics so emqtt-bench now supports publish to different topics. This is new{publish, Topic}method which means some CLI args for publishing are not used.This branch has its value for testing the QUIC multi streams that maps different priority topics onto different streams to mitigate the HOL blocking issue.
this branch deps on: emqx/emqtt#252