[
https://issues.apache.org/jira/browse/LOG4J2-3115?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Markus Spann updated LOG4J2-3115:
---------------------------------
Description:
Implementations of {{NamePatternConverter}} are responsible for abbreviating
logger and class names according to a configured pattern.
Abbreviation can be quite complex and computationally expensive (consider
abbreviating x parts of a fully qualified class name to y number of leftmost
characters).
Example: The pattern {{%logger{1.}}} abbreviates logger name
{{org.apache.logging.log4j.core.pattern.NameAbbreviator}} to
{{o.a.l.l.c.p.NameAbbreviator}}
The strings to abbreviate (logger and class names) extremely likely stem from a
limited set of candidates. I would like to suggest that results are cached to
prevent computation over and over again e.g. in every log statement of a logger.
I have a small patch ready that addresses the issue.
was:
Implementations of {{NamePatternConverter}} are responsible for abbreviating
logger and class names according to a configured pattern.
Abbreviation can be quite complex and computationally expensive (consider
abbreviating x parts of a fully qualified class name to y number of leftmost
characters).
Example: The pattern {{%logger\{1.}}} abbreviates logger name
{{org.apache.logging.log4j.core.pattern.NameAbbreviator}} to
{{o.a.l.l.c.p.NameAbbreviator}}
The strings to abbreviate (logger and class names) are extremely likely stem
from a limited set of candidates. I would like suggest that results are cached
in a map to prevent computation over and over again e.g. in every log statement
of a logger.
I have a small patch ready that addresses the issue.
> Remember previously computed strings in NamePatternConverter
> ------------------------------------------------------------
>
> Key: LOG4J2-3115
> URL: https://issues.apache.org/jira/browse/LOG4J2-3115
> Project: Log4j 2
> Issue Type: Improvement
> Components: Core
> Reporter: Markus Spann
> Priority: Minor
>
> Implementations of {{NamePatternConverter}} are responsible for abbreviating
> logger and class names according to a configured pattern.
> Abbreviation can be quite complex and computationally expensive (consider
> abbreviating x parts of a fully qualified class name to y number of leftmost
> characters).
> Example: The pattern {{%logger{1.}}} abbreviates logger name
> {{org.apache.logging.log4j.core.pattern.NameAbbreviator}} to
> {{o.a.l.l.c.p.NameAbbreviator}}
> The strings to abbreviate (logger and class names) extremely likely stem from
> a limited set of candidates. I would like to suggest that results are cached
> to prevent computation over and over again e.g. in every log statement of a
> logger.
> I have a small patch ready that addresses the issue.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
