[
https://issues.apache.org/jira/browse/YARN-11429?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17732516#comment-17732516
]
ASF GitHub Bot commented on YARN-11429:
---------------------------------------
brumi1024 commented on PR #5736:
URL: https://github.com/apache/hadoop/pull/5736#issuecomment-1591114150
Thanks @tomicooler for the update, looks good to me, and improves the
readability.
It's not strictly related to this patch, and probably not in its scope, but
I don't really get why are we doing the following flow for a simple JSON pretty
print: string -> JSON POJO (lib#1) -> string -> create an ObjectMapper per
pretty print (it's a really heavyweight object, it should be reused) -> JSON
POJO (lib#2) -> string again. This is much-much slower than necessary. Probably
this should be refactored later.
Regardless I'll merge this now, and create a followup ticket.
> Improve updateTestDataAutomatically in TestRMWebServicesCapacitySched
> ---------------------------------------------------------------------
>
> Key: YARN-11429
> URL: https://issues.apache.org/jira/browse/YARN-11429
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: yarn
> Reporter: Tamas Domok
> Assignee: Tamas Domok
> Priority: Major
> Labels: pull-request-available
>
> The
> [updateTestDataAutomatically|https://github.com/apache/hadoop/blob/bfce21ee08f2faef795354ac3615eb2a393bd158/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/webapp/TestRMWebServicesCapacitySched.java#L340]
> helper function should have a fix json format to be useful.
> {code:java}
> String actual = json.toString(2);
> updateTestDataAutomatically(expectedResourceFilename, actual);
> assertEquals(
> prettyPrintJson(getResourceAsString(expectedResourceFilename)),
> prettyPrintJson(actual));
> {code}
> The comparison is working with different json formats which is all right, but
> updating the test data should be easy, and the git diff should be easily
> readable by humans.
> Something like this could work:
> {code}
> String actual = prettyPrintJson(json.toString(2));
> updateTestDataAutomatically(expectedResourceFilename, actual);
> assertEquals(
> getResourceAsString(expectedResourceFilename),
> actual);
> {code}
> The json's must be updated unfortunately with the new format.
> The defaults shouldn't change here. Maybe we should be even more explicit
> with the format options:
> {code}
> private static String prettyPrintJson(String in) throws
> JsonProcessingException {
> ObjectMapper objectMapper = new ObjectMapper();
> return objectMapper
> .writerWithDefaultPrettyPrinter()
> .writeValueAsString(objectMapper.readTree(in));
> }
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
