use native json for stabler parsing feature

kengz / SLM-Lab

Modular Deep Reinforcement Learning framework in PyTorch. Companion library of the book "Foundations of Deep Reinforcement Learning".

https://slm-lab.gitbook.io/slm-lab/

MIT License

1.23k stars 263 forks source link

use native json for stabler parsing feature #486

Closed Nickfagiano closed 3 years ago

Nickfagiano commented 3 years ago

details filled in by @kengz

Some users experience JSON parsing error caused by a newer version of ujson. This replaces the ujson call when parsing spec files with the native json package for better stability.

kengz commented 3 years ago

CI unit test failures shows that ujson is still needed. Best course of action is to:

use strictly ujson=1.35. If you manually installed ujson and it picked up a newer version, correct it with conda install ujson=1.35
manually hunt down and correct the extra comma error in spec files