It's hard to provide atomic operations for every resources, like in relational databases. For example, if a tool writes to a file, like unzip, it constructs the file steps by step.
Another approch is to provide an API where you can know if a resource is usable. When a new resource has been produced, when it won't change anymore,
the state of this resource could change to available. With this system, it's the client responsability not to use a resource that is not fully availlable yet.
Any other idea to tend towards atomicity is welcome...
It's hard to provide atomic operations for every resources, like in relational databases. For example, if a tool writes to a file, like
unzip, it constructs the file steps by step.Another approch is to provide an API where you can know if a resource is usable. When a new resource has been produced, when it won't change anymore, the state of this resource could change to available. With this system, it's the client responsability not to use a resource that is not fully availlable yet.
Any other idea to tend towards atomicity is welcome...