Is it actually possible to create a fixed-size looping buffer? about mlib

[P-p-H-d]

Yes, it is possible. It is indeed m-buffer (BUFFER_DEF with BUFFER_UNSAFE|BUFFER_QUEUE)
Why do you think it is not suitable?

[vladipus]

I would imagine that a ring fixed-size buffer is something that loops its new last element all over again. Your implementation just rewrites the count - 1 element when the buffer is full.

[P-p-H-d]

I am not sure to understand what you mean.
BUFFER_DEF with a BUFFER_QUEUE implements a ring buffer as explained here: https://en.wikipedia.org/wiki/Circular_buffer

[vladipus]

Does start and end pointers actually change in your implementation?

[P-p-H-d]

Yes

[vladipus]

What happens when you push an element to a full queue?
I want a thread-unsafe ring-buffer and using BUFFER_PUSH_OVERWRITE also. The pushing seems to always overwrite the last element (count - 1)

[vladipus]

I seem to miss the point, maybe. is BUFFER_PUSH_OVERWRITE mandatory for thread-unsafe buffers?

[P-p-H-d]

What happens when you push an element to a full queue?

It depends on the configuration of the queue. It may:

• block until the queue is not full (if BUFFER_THREAD_SAFE)
• overwrite the youngest element if BUFFER_PUSH_OVERWRITE
• return a failure otherwise.

I want a thread-unsafe ring-buffer and using BUFFER_PUSH_OVERWRITE also. The pushing seems to always overwrite the last element (count - 1)

This is what BUFFER_PUSH_OVERWRITE does (overwrite the youngest element in case of buffer full) to avoid blocking.

is BUFFER_PUSH_OVERWRITE mandatory for thread-unsafe buffers?

No

[vladipus]

Ok, I now seem to get it. Thank you.
But it still looks kinda un-optimal that atomic is used in a thread-unsafe version.
what do you think?

[P-p-H-d]

It is true! Thanks. I have opened an issue on it: #90