search

eclipse-zenoh / zenoh-c

C API for Zenoh
http://zenoh.io
Other
82 stars 57 forks source link

[Bug] The SHM buffer check error log is misleading #814

Open YuanYuYuan opened 2 weeks ago

YuanYuYuan commented 2 weeks ago

Describe the bug

When sending SHM buffers across machines, we check the SHM buffer availability on the receiver side with this function. However, this reports an error that is NOT shown in zenoh rust example. Since we had already properly returned the result in the zenoh-c function, do we need to duplicate this error in the log?

To reproduce

  1. Build the examples _z_pubshm and _z_subshm.
  2. Run the two examples on different hosts.
  3. The error log is shown below. 
    2024-11-12T06:35:36.001040Z  INFO ThreadId(02) zenoh::net::runtime::orchestrator: Zenoh can be reached at: tcp/192.168.0.109:34133
2024-11-12T06:35:36.001490Z  INFO ThreadId(02) zenoh::net::runtime::orchestrator: zenohd listening scout messages on 224.0.0.224:7446
Declaring Subscriber on 'demo/example/**'...
Enter 'q' to quit...
2024-11-12T06:35:36.176089Z ERROR rx-1 ThreadId(07) zenohcd::zbytes: Failed to convert the payload
>> [Subscriber] Received PUT ('demo/example/zenoh-rs-pub': '[  91] Pub from Rust SHM!') [RAW]
2024-11-12T06:35:37.175526Z ERROR rx-1 ThreadId(07) zenohcd::zbytes: Failed to convert the payload
>> [Subscriber] Received PUT ('demo/example/zenoh-rs-pub': '[  92] Pub from Rust SHM!') [RAW]

System info

yellowhatter commented 1 week ago

@YuanYuYuan I think we are using this: https://github.com/eclipse-zenoh/zenoh-c/blob/153ad8dc5b17d874ded973cc62dddbe56eefa746/src/zbytes.rs?plain=1#L188

And yes, IMHO we should:

  1. Rename all as_shm* to try_as_shm* in API
  2. Do not post error messages for try_as_shm

@milyin @DenisBiryukov91 what do you think?

yellowhatter commented 1 week ago

In opposite to other as_* like as_string, as_shm is used to probe the buffer on shm nature, and unsuccessful result is not as hard error as for orther APIs