[
https://issues.apache.org/jira/browse/HTTPCLIENT-2096?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Oleg Kalnichevski resolved HTTPCLIENT-2096.
-------------------------------------------
Resolution: Fixed
> Consider using static loggers instead of instance loggers
> ---------------------------------------------------------
>
> Key: HTTPCLIENT-2096
> URL: https://issues.apache.org/jira/browse/HTTPCLIENT-2096
> Project: HttpComponents HttpClient
> Issue Type: Bug
> Affects Versions: 5.0.1
> Reporter: Carter Kozak
> Priority: Major
> Time Spent: 2h 10m
> Remaining Estimate: 0h
>
> I'd like to consider replacing instance loggers with static loggers to avoid
> paying a penalty on instantiation. This would result in slightly different
> (and arguably better) behavior when classes are extended, as the logger
> origin would reflect the location that produced logging, not the subclass
> that happened to extend it. For example, if I want to debug hc5, I might set
> the {{org.apache.hc}} origin to {{TRACE}} level and add it to a separate
> appender. If I've extended a component it won't appear with my httpclient
> logs using the {{getClass()}} approach.
> From [~olegk]
> bq. As far as I know static loggers in low level libraries are considered
> evil and are to be avoided in favor of instance loggers. The overhead of
> logger instantiation is quite manageable if one injects logger through a
> constructor of frequently allocated and deallocated components.
> Conversation started here:
> https://github.com/apache/httpcomponents-client/pull/232#discussion_r448963006
> Some existing documentation:
> http://www.slf4j.org/faq.html#declared_static
> https://logging.apache.org/log4j/2.x/manual/usage.html#StaticVsNonStatic
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
