18 Mar
2024
18 Mar
'24
1:51 p.m.
On Mon Mar 18 13:34:54 2024 +0000, Jinoh Kang wrote:
What's your thinking of removing id from `session_shm_t` altogether? It's currently 0, which directly contracts with
object unique id, object data is valid if != 0
Imo it makes it more awkward to use the shared read/write macros. They assume an object with its header to be passed, and using them for the session implies that it *has* to be an object. Faking it by using a different struct with just a seq feels wrong. If the session being an "invalid" object feels too awkward I'd prefer to simply set its id to be non-zero instead.