The accelerated stream processing library has an abstraction for a specific (local) data store . I looked at the source code, and it seems that the TSS function can be used in the application with any existing thread, no matter what the weather was created from boost :: thread --ie, this means that certain callbacks are registered with the kernel before intercepting the callback function a call that can call the destructor of any TSS objects when a thread or process goes out of scope. I found these callbacks.
I need to cache HMAC_CTX from OpenSSL inside the workflows of various web servers (see this , detailed, question about what I'm trying to do), and therefore I do not control the lifetime of the stream - the web server. Therefore, I will use the TSS functionality for threads not created by boost :: thread.
I just wanted to test my assumptions before starting to implement the caching logic, are there any flaws in my logic?
source
share