[
https://issues.apache.org/jira/browse/FELIX-6664?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17779565#comment-17779565
]
Dominik Süß commented on FELIX-6664:
------------------------------------
created fixed PR at https://github.com/apache/felix-dev/pull/241
I now took care to consider comments crossing multiple reads - to enable using
bufferedReader without adding too much complexity I changed the newline
behavior slightly - any newlines within a comment are dropped (but it's not
dropping the closing newline and leading spaces, only the full comments start
to end)
> Comment Removing on JSONSupport MemoryInefficient
> -------------------------------------------------
>
> Key: FELIX-6664
> URL: https://issues.apache.org/jira/browse/FELIX-6664
> Project: Felix
> Issue Type: Improvement
> Reporter: Dominik Süß
> Priority: Major
>
> While using JSONSupport in Context of the Sling Featurelauncher we recently
> stumbled over an OOM exception related to how JSONSupport handles removal of
> comments:
> {code}
> java.lang.OutOfMemoryError: Java heap space
> at java.base/java.util.Arrays.copyOfRange(Arrays.java:4030)
> at java.base/java.lang.StringUTF16.newString(StringUTF16.java:1025)
> at java.base/java.lang.StringBuilder.toString(StringBuilder.java:454)
> at
> org.apache.felix.cm.json.io.impl.JsonSupport.removeComments(JsonSupport.java:308)
> at
> org.apache.felix.cm.json.io.impl.JsonSupport.createCommentRemovingReader(JsonSupport.java:244)
> at
> org.apache.felix.cm.json.io.Configurations.jsonCommentAwareReader(Configurations.java:70)
> at
> org.apache.sling.feature.io.json.FeatureJSONReader.readFeature(FeatureJSONReader.java:676)
> {code}
> https://github.com/apache/felix-dev/blob/91432d1a3f08520d5eb75b5c8e3443bb75f7c467/cm.json/src/main/java/org/apache/felix/cm/json/io/impl/JsonSupport.java#L233-L257
> The code does use a StringWriter to create a full String representation of
> the featuremodel and then acts on that model. As this featuremodel can
> contain a lot of metadata in comments those can get a significant size and
> when being used in a resource constrained environment can lead to memory
> issues.
> I prepared a patch that doesn't touch the removal logic but simply creates a
> custom BufferedReader that performs the removal on read and therefore
> eliminates the most prominent area for resource optimization.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
