Is Deadpool still simple?

[Qqwy]

The README states the following:

Deadpool is simple. Dead simple. There is very little API surface. The actual code is barely 100 lines of code and lives in the two functions Pool::get and Object::drop.

However, using cloc on the src/ directory gives the following results:

-------------------------------------------------------------------------------
Language                     files          blank        comment           code
-------------------------------------------------------------------------------
Rust                            12            210            479           1326
-------------------------------------------------------------------------------
SUM:                            12            210            479           1326
-------------------------------------------------------------------------------

Even if only considering the unmanaged part of Deadpool, we get:

-------------------------------------------------------------------------------
Language                     files          blank        comment           code
-------------------------------------------------------------------------------
Rust                             3             46            136            313
-------------------------------------------------------------------------------
SUM:                             3             46            136            313
-------------------------------------------------------------------------------

So can Deadpool still be considered simple?
At the very least, the 'barely 100 lines' statement no longer seems true.

Or am I misunderstanding the statement?

[bikeshedder]

Deadpool used to be "dead simple". Over the time features were added: Timeouts, hooks, metrics, etc.

My goal has always been to stay true to the original statement but some compromises had to be made...

The core logic I was refering to lives in timeout_get (50 LoC):

https://github.com/bikeshedder/deadpool/blob/387f9e25d2d6197aab4ed6e3218cc6d29a2b35f9/src/managed/mod.rs#L321-L376

A while ago try_recycle ...

https://github.com/bikeshedder/deadpool/blob/387f9e25d2d6197aab4ed6e3218cc6d29a2b35f9/src/managed/mod.rs#L379-L421

... and try_create were extracted from the timeout_get function to clean up the code a bit:

https://github.com/bikeshedder/deadpool/blob/387f9e25d2d6197aab4ed6e3218cc6d29a2b35f9/src/managed/mod.rs#L424-L456

You're right though. 100 LoC is a massive understatement. There are lots more lines of code that make up the pool implementation nowadays.

[Qqwy]

Thank you very much for the clarification!

[bikeshedder]

btw. the config.rs file alone is 147 LoC. It is code. But is it also part of the logic? It's some simple structures that can be (de)serialized and implement some useful traits. metrics.rs and errors.rs are similar in that regards. Very little actual "program logic" but still need to be considered "code".

I do agree that the current implementation has gotten a bit out of hands with the differentiation between UnreadyObject and InnerObject. If #336 ever lands (sth. I still haven't made up my mind 100%, yet) the code needs some massive changes anyways and I might get away removing a bulk of special logic and even simplify the code a bit. Though bringing the size down to "true" 100 LoC is simply impossible as the config code alone is bigger in size. 😅

Maybe that statement can be reworked a bit? Calling it "dead simple" really is a massive exaggeration. I originally went with it because of the name. The whole reason for naming this project "deadpool" was because of that pun. 🙈