[
https://issues.apache.org/jira/browse/OLINGO-184?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13917877#comment-13917877
]
Stephan Klevenz commented on OLINGO-184:
----------------------------------------
Have a look at
https://git-wip-us.apache.org/repos/asf?p=incubator-olingo-odata4.git;a=blob;f=odata4-lib/odata4-commons-core/src/main/java/org/apache/olingo/odata4/commons/core/edm/primitivetype/EdmDouble.java;h=743802d4c717935b406410b530f446c5a61dd914;hb=HEAD
The regex which is used to verify the validity of a double is 4 times slower
than without it. And nearly the same rule is applyed within the double
conversion. If an OData double into an Java double we just should accept it.
Of course, rounding gets a problem. What you put in is not what you'll get out.
But the regex not solves that too.
> Performance optimization on edm primitve types
> ----------------------------------------------
>
> Key: OLINGO-184
> URL: https://issues.apache.org/jira/browse/OLINGO-184
> Project: Olingo
> Issue Type: Improvement
> Components: odata4-commons
> Reporter: Stephan Klevenz
> Priority: Minor
>
> Edm primitive types are doing a lot of reg ex checks to be 100% compliant
> with OData 4.0 specification. These checks are very expensive in CPU
> consumption.
> Maybe it is possible to avoid this checks (and instead of OData rules the
> Java type system rule will take place) or if that is not possible make them
> optional.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
