Phoenix: Investigate using Jason instead of Poison
This will also require changing phoenix_ecto / ecto.
josevalim
All 9 comments
What can of investigation needs to be done for this? Is it:
- How much effort it would be
- Seeing if there is a performance boost
- Something else?
wilbertom
on 6 Jan 2018
It won't be a large effort. The only thing to evaluate is the readiness of Jason stability and prod ready wise.
chrismccord
on 7 Jan 2018
The biggest difference of Jason vs Poison is that Jason does not allow encoding arbitrary structs - it always requires either implementing or deriving the protocol. This would mean the change wouldn't be completely seamless - fortunately, deriving the protocol is just one line Protocol.derive(Jason.Encoder, MyStruct).
michalmuskala
on 7 Jan 2018
Just an FYI, I've released version 1.0.
michalmuskala
on 26 Jan 2018
@josevalim this is taken care of in #2734 , and phoenix_ecto no longer depends or implements any Poison protocol so we are good there. What needs done on Ecto's side to make this complete?
chrismccord
on 30 Jan 2018
Can someone explain what are the benefits of using Jason instead of Poison?
mpranjic
on 14 May 2018
The primary reason is performance. Jason is about twice as fast and uses half the memory while being almost 100% functionally compatible with Poison (it covers all the features used directly in Phoenix).
michalmuskala
on 14 May 2018
Will this be released to 1.3?
javierg
on 24 Jul 2018
@javierg if you look at Jason's readme, instructions are provided for use with phoenix 1.3:
https://github.com/michalmuskala/jason/blob/master/README.md
beardedeagle
on 24 Jul 2018
Related issues
bwschmidt
路
4Comments
Tica2
路
3Comments
noam87
路
3Comments
fertapric
路
3Comments
mitchellhenke
路
3Comments
Most helpful comment
The primary reason is performance. Jason is about twice as fast and uses half the memory while being almost 100% functionally compatible with Poison (it covers all the features used directly in Phoenix).