is it the expectation that atomic should be used for all class level variables now?

#crystallang

@toddsundsted I don’t thinks so. Why? If this is coming from parallel unsafety sharing there are more things to worry also: like closures union type values accessed from multiple threads.

@bcardiff that was me thinking out loud.

i'm running code checks on ktistec looking for possible issues. the checks identified class variable read-modify-write patterns in code meant to run in concurrent fibers as possible future multithreading issues and suggested using atomic as a mitigation now.

that made me wonder if this was a pattern i should be thinking about when writing concurrent code that accesses shared state in class variables (even if i'm only using fibers now).