Open CEHENKLE opened 2 years ago
peternied
commented
2 years ago Common Utils has presented with ownership hiccups with everyone and no-one being responsible directly for it. I would prefer the clarity of putting it inside the more actively managed codebase
nknize
commented
2 years ago We have not revisited whether this still makes sense in an OpenSearch world :)
Seems premature. I think this :point_up: should be resolved before any refactoring to core is done?
dblock
commented
2 years ago We have not revisited whether this still makes sense in an OpenSearch world :)
Seems premature. I think this ☝️ should be resolved before any refactoring to core is done?
It seems like merging it into core is a way to assign ownership. This may or may not be bad.
@nknize Do you have technical arguments not to?
peternied
commented
2 years ago We have not revisited whether this still makes sense in an OpenSearch world :)
CommonUtils was a mechanism for all proto-OpenSearch codebases to shared code. In the era of OpenSearch, the utility of this code as a separate library has been minimized.
If someone proposed making a CommonUtils2.0 repository today I would vote against its creation.
nknize
commented
2 years ago @nknize Do you have technical arguments not to?
Posted at https://github.com/opensearch-project/OpenSearch/issues/2545#issuecomment-1076378627
:-1: for 2.0.0 it won't be ready:
"Enable this after integration with security plugin is done" So why aren't we just integrating the appropriate classes with the security plugin?:+1: for a future release if targeting core makes sense. We might want to integrate with security first (or dissolve the implementations across other plugins) then integrate security with core rather than blindly integrate with core.
CEHENKLE
commented
2 years ago @praveensameneni Can you share on this issue (or point us to another issue) your analysis of the state of common utils and your plan to clean up the issues @nknize pointed out?
Thanks!
CEHENKLE
commented
2 years ago @praveensameneni Hi, I'd like to pick this back up and dust it off. I think folding common utils into core is the right direction, but I don't want to do that until the items Nick mentioned are done. Are you actively working on that, and if so, what's the ETA?
praveensameneni
commented
2 years ago @CEHENKLE , Just some background, we created common-utils for supporting fine grained access in different plugins which has been adopted by many plugins - alerting, Anomaly Detection, Index Management, Async search among others. Over time few other plugins added code into common-utils. In addition to the items that @nknize posted, it needs to be re-designed and this will impact several plugins. We are not actively working on the re-design of common-utils.
dblock
commented
2 years ago I must point out that this existential issue was written on February 11th, which is the 42nd day of the year.
raj-chak
commented
1 year ago
Is your feature request related to a problem? Please describe. Common Utils was stood up as an independent repo back in the days of ODFE. We have not revisited whether this still makes sense in an OpenSearch world :)
This issue is to evaluate if it makes sense to move common utils into core, and if so how/where.
Please note: Common Utils right now is a grab bag of different features and functions, many of which are marked deprecated. We'd want to make sure we were moving over only the things that made sense in a mindful manner, rather than a straight up "lift and shift" as is.
Require work: Accounting and evaluation of what's currently in common utils. For each item, determining the best location to move it into core (or into another plug in if applicable).