Why not sync.Pool

[nexovec]

Hi, I realize this is a noob question, but why shouldn't I just use sync.Pool for a resource pool?

[ccoVeille]

It's a good question.

I think the README should be updated with information that might be shared here

[jackc]

They are intended for entirely different types of resources. sync.Pool would generally be used for in memory objects. Puddle would generally be used for handles to external objects such as connections, file handles, etc.

• sync.Pool does not have a way to limit max resources in pool.
• sync.Pool does not have a way to ensure at least min resources in pool.
• sync.Pool can drop resources in pool at any time - not ideal for expensive to create connections.
• sync.Pool does not have a cleanup / release function. Resources are GCed without a chance to close cleanly.
• sync.Pool does not have a way to handle errors creating the resource
• sync.Pool does not support context to limit time to wait creating a resource

[ccoVeille]

thanks for your valuable and detailed reply, I opened a PR to complete the README with these elements

Feel free to amend it

[nexovec]

I'm looking specifically for something that would help me reuse byte slices and other such objects to ease gc pressure. @jackc 's response makes it seem like that's not the intended use case for this library. Am I in the wrong for trying to use it this way?

It's not like I need to do this somewhere, but I do know the lifetimes of my byte slices most of the time, so why not reuse them(mostly as some sort of pseudo arena allocator)?

[jackc]

@nexovec Easing GC pressure is the raison d'etre for sync.Pool. You almost certainly want to use that over puddle.