diff options
Diffstat (limited to 'include/NT.h')
| -rw-r--r-- | include/NT.h | 202 |
1 files changed, 202 insertions, 0 deletions
diff --git a/include/NT.h b/include/NT.h new file mode 100644 index 0000000..a24d9d4 --- /dev/null +++ b/include/NT.h @@ -0,0 +1,202 @@ +#pragma once +#include <cstdint> +#include <set> +#include <vector> +#include <string> + +#if defined(_WIN32) && defined(NT_CONSOLE_APP) +#define NT_API +#elif defined(_WIN32) && defined(NT_EXPORTS) +#define NT_API __declspec(dllexport) +#elif defined(_WIN32) +#define NT_API __declspec(dllimport) +#else +#define NT_API +#endif + +enum OBJECT_TYPE { + MULTIGROUP, + RELATION, + TUPLE, + ATTRIBUTE, + TRANSACTION +}; + +enum METHOD { + OPEN, + CLOSE, + PARSE, // To plug a secondary namespace + SECURITY +}; + +enum AUTH_CLAIM { + READ, + WRITE +}; + +enum AUTH_METHOD { + CERTIFICATE, + PLAIN_TEXT +}; + +namespace nt +{ + class NT_API ObjectManager + { + public: + class IObject { + public: + virtual ~IObject() = default; + }; + class Multigroup : IObject {}; + class Relation : IObject {}; + class Tuple : IObject {}; + class Transaction : IObject {}; + struct object_type { + OBJECT_TYPE label; + // If the object is not durable. E.g.: TRANSACTION + bool disposable; + std::set<METHOD> methods; + }; + struct registry_head { + uint32_t reference_count; + uint32_t handle_count; + struct object_type* type; + std::vector<std::string> path; + //security_descriptor to be added later, maybe also a set of labels + }; + struct registry { + struct registry_head* head; + IObject object; + // TODO: Treat as a list for now, but ideally it's a tree + struct registry* next; + }; + struct registry** entries; + // Looks into this registry and attempts to retrieve the entry + struct registry* Find(const std::vector<std::string> object_path) { return nullptr; } + }; + + class NT_API PermissionsManager + { + public: + // For validating connections at login + std::set<AUTH_CLAIM> Firewall(AUTH_METHOD method) { return {}; } + // For validating permission to an object + const bool Access(const ObjectManager::registry* object, const void* connection_context) { + // Checks the security descriptor on the head of the object we find in the ObjectManager::registry + // Then additionally check other metadata yet to be defined in the connection_context, + // like user group, policy, etc. + // Yet to decide if this should be summoned at every attempt to get ahold of an object, + // that might produce significant overhead. How can we solve this? + return object != nullptr && connection_context != nullptr; + } + }; + + class NT_API NamespaceReferenceManager + { + }; + + class NT_API CursorManager + { + }; + + class NT_API IdentityManager + { + public: + // Looks into the methods of the object type and asserts if it can be opened + const bool CanOpen(ObjectManager::registry* object) { return object != nullptr; } + const bool CanClose(ObjectManager::registry* object) { return object != nullptr; } + }; + + class NT_API LifecycleManager + { + public: + // If object is not being monitored already, monitors the lifecycle by updating the handle count. + // Maybe also look at the reference count and kill some categories of disposable objects no longer referenced by other objects, like transactions. + // Think of other use cases later. + void Monitor(ObjectManager::registry* object) {} + // Serializes contention for the change of a mutable reference. + // This can be used in, for example, the HEAD of a branch to serialize the access. + // If the object does not causes contention, like a relation (since they are immutable, just like multigroups, all snapshots) + // then just approve, as they do not cause any contention. Transactions also do not cause any, + // the only things that do are mutable reference states in shared sessions, + // for example: the HEAD of a branch, namespace collisions (creating two branches with the same name, or two multigroups with the same name from the same state, etc), + // multi-branch atomic commits (which I wish to not have as it damages isolation), + // garbage collection of old state (as in, we want to delete an old-enough history to save up space), etc. + const bool Contention(ObjectManager::registry* object){ return object != nullptr; } + // The inverse of `Monitor`, stops monitoring. + // Maybe garbage collect here also if the counters goes to zero. + void Unmonitor(ObjectManager::registry* object) {} + }; + + class NT_API HandlerManager + { + public: + // This is supposed to represent an authorized access to a certain object in the registry + // This increases the handle count in the object manager + // We should consider adding on the object manager a vector of all the handles that refer to an object + struct handle { + ObjectManager::registry* object = nullptr; + void* connection_context = nullptr; + }; + void DeallocateHandle(struct handle* handle) {} + struct handle Open(std::vector<std::string> object_path, void* connection_context) { + ObjectManager objects; + PermissionsManager permissions; + IdentityManager identities; + LifecycleManager lifecycles; + ObjectManager::registry* retrieved_object = objects.Find((const std::vector<std::string>)object_path); + if (permissions.Access(retrieved_object, (const void*)connection_context) + && identities.CanOpen(retrieved_object) + && lifecycles.Contention(retrieved_object)) { + // Starts monitoring this guy + lifecycles.Monitor(retrieved_object); + return { retrieved_object, connection_context }; // Return a successful handle + } + else { + // Return something more algebraic instead of null or an int + return {}; + }; + } + bool Close(struct handle *handle) { + if (handle == nullptr) { + return false; + } + PermissionsManager permissions; + IdentityManager identities; + LifecycleManager lifecycles; + // Somehow grab the object pointer and connection (maybe drop it...) from the handle + ObjectManager::registry* object = handle->object; + void* connection_context = handle->connection_context; + // Checking for permissions here may be relevant. + // What if we have a handle but no permission? Spooky + if (permissions.Access(object, connection_context) + && identities.CanClose(object) + // TODO: I think that in this case we do not need to worry about contention. + // But be careful to evaluate this scenario later. + //&& LifecycleManager::Contention(retrieved_object) + ) { + // Stops monitoring this guy + lifecycles.Unmonitor(object); + DeallocateHandle(handle); + return true; // Return a successful handle + } + else { + // Return something more algebraic instead of null or an int. + // It's kinda bad to not even report what went wrong. + return false; + }; + } + }; + + class NT_API WAM + { + }; + + class NT_API NT + { + public: + bool IsRunning() const; + void SimulateEntryCall() {}; + }; +} |