Which candidate to test transport layer ?

[sbernard31]

To test abstraction Layer, it could make sense to try to implement a new transport based on something else than Californium.

One possible way is to create a new CoAP endpoint Provider based another CoAP java library.

I tried to look what is currently available 👇

name	repository	last mod.	feature	maven central	alt. maven repo
java-coap (fork)	https://github.com/open-coap/java-coap	~ some days ago	observe, block transfer, coap+tcp, coaps+tcp, coaps (based on mbedtls)	✔️
java-coap	https://github.com/PelionIoT/java-coap	~ 1, 2 years ago	observe, block transfer, coap+tcp, coaps+tcp	❌	✔️
ncoap	https://github.com/okleine/nCoAP	~ 5 years ago	observe, block transfer	❌	✔️
coap-blaster	https://github.com/google/coapblaster	~ 3, 4 years ago	observe, block2	✔️	❌
jcoap	https://gitlab.amd.e-technik.uni-rostock.de/ws4d/jcoap	~ 3/4 years ago	?	❌	❌
mjCoAP	❌	?	observe, block transfer	❌	❌

I also find there was a project to provide a CoAP implementation in OpenJDK known as Kona but it seems the project is archived.
Too bad, It could have been a simple default implementation without any dependencies and californium would have been the full feature one.

[sbernard31]

Another solution is to directly target another protocol like experiment about mqtt ? #1236

[sbernard31]

If we decide to test a new Java CoAP implementation, for now the better choice would be java-coap (fork).

[sbernard31]

Discussion about current experimentation with java-coap(fork) at #1373

[sbernard31]

We provide now a CoAP implementation based on java-coap(fork), See #1373 for more details.

I think we can close this issue.