use of var in a low-level library. The lower you go, the more explicit types should be. Why introduce another potential source of errors? Can we remove var and replace with class names?
use of streams/lambdas/collecting and in general anywhere we are now creating new objects when we could have avoided it. low level libraries should generally prefer for-loops and static classes. Can you review and avoid using stream methods and allocating new objects where possible?
Please give me a heads-up in advance before refactoring more library code. With the above changes I will merge this, but frankly I don’t see a big benefit given the risk. If we had more comprehensive test coverage I would feel better about the risk.
@kris -- Things that make me wary:
* use of `var` in a low-level library. The lower you go, the more explicit types should be. Why introduce another potential source of errors? Can we remove `var` and replace with class names?
* use of streams/lambdas/collecting and in general anywhere we are now creating new objects when we could have avoided it. low level libraries should generally prefer for-loops and static classes. Can you review and avoid using stream methods and allocating new objects where possible?
Please give me a heads-up in advance before refactoring more library code. With the above changes I will merge this, but frankly I don't see a big benefit given the risk. If we had more comprehensive test coverage I would feel better about the risk.
Removing var (only String was replaced in 5ish places, so it was easy) - agree, sorry
stream/lambda/collect is used just in 1 place - method prefix(Collection<String>). In master (current version) following is used transform(keys, o -> prefix(o.toString())) - so lambda is there also. I did some perfomance test on this, and the new solution worked 2x faster at least on collections with 10.000 Strings. Please confirm you really want apache’s transform method back. Or maybe to just change all this and use plain for loop with another collection created as output?
- Removing var (only String was replaced in 5ish places, so it was easy) - agree, sorry
- stream/lambda/collect is used just in 1 place - method `prefix(Collection<String>)`. In master (current version) following is used `transform(keys, o -> prefix(o.toString()))` - so lambda is there also. I did some perfomance test on this, and the new solution worked 2x faster at least on collections with 10.000 Strings. Please confirm you really want apache's transform method back. Or maybe to just change all this and use plain for loop with another collection created as output?
jonathan
WIP: (testing) Cleaning redis service为 Cleaning redis service adn adding ttl getter method@jonathan please review
Cleaning redis service adn adding ttl getter method为 Cleaning redis service and adding ttl getter method@kris -- Things that make me wary:
varin a low-level library. The lower you go, the more explicit types should be. Why introduce another potential source of errors? Can we removevarand replace with class names?Please give me a heads-up in advance before refactoring more library code. With the above changes I will merge this, but frankly I don’t see a big benefit given the risk. If we had more comprehensive test coverage I would feel better about the risk.
prefix(Collection<String>). In master (current version) following is usedtransform(keys, o -> prefix(o.toString()))- so lambda is there also. I did some perfomance test on this, and the new solution worked 2x faster at least on collections with 10.000 Strings. Please confirm you really want apache’s transform method back. Or maybe to just change all this and use plain for loop with another collection created as output?5d199cbd98被合并。