Feature comparison matrix

[rossabaker]

I have to relearn this every time there's an issue, so I'm writing it down. Presented in hopes it guides feature development, inspires documentation, and assists any translations.

Commons-Pool2	Parity?	keypool-0.4.7	keypool-0.4.8	blaze	skunk
blockWhenExhausted	#555	never exhausted	true (see #555)	true 1	true
durationBetweenEvictionRuns	#556	5.seconds	5.seconds	2
evictionPolicy	#558	f(idleTimeAllowedInPool)	f(idleTimeAllowedInPool)
evictorShutdownTimeoutDuration
fairness	#559	never exhausted	false	true	true
jmxEnabled		3	3
lifo		false	false	true	true
logAbandoned
maxIdlePerKey		maxPerKey 3	maxPerKey 3		no keys
maxTotalPerKey				maxConnectionsPerRequestKey	no keys
maxTotal			maxTotal 4	maxTotal	size
maxWaitDuration		5	5	5	5
minEvictableIdleDuration		idleTimeAllowedInPool	idleTimeAllowedInPool
minIdlePerKey					no keys
numTestsPerEvictionRun
removeAbandonedOnBorrow
removeAbandonedOnMaintenance
removeAbandonedTimeoutDuration
softMinEvictableIdleDuration
testOnBorrow		false	false	true 6	false
testOnCreate		false	false	false	false
testOnReturn		false	false	false	true 7
testWhileIdle		false	false	false	false
				maxWaitQueueLimit
		idleTimeAllowedInPool	idleTimeAllowedInPool	maxIdleDuration
		maxTotal

Footnotes

1. evicts a random idle connection ↩

2. evicts on borrow ↩

3. Draft in Measure metrics using otel4s #396 ↩ ↩² ↩³ ↩⁴

4. Buggy: MaxTotal does not take idle resources into account #440 ↩

5. Could be expressed with .timeoutTo ↩ ↩² ↩³ ↩⁴

6. hardcoded to Blaze connections ↩

7. with recycler ↩

[rossabaker]

I'm creating a commons-pool2 parity label to contemplate improvements. Full parity is not a goal, but an understanding of where we're forgoing parity and why is.