Julian Reschke commented on OAK-7335:

 - no java character takes more than *3* bytes in UTF-8, due to the fact that 
plane 1 unicode characters are represented as surrogate pairs (thus 2 Java 
characters). That said, I'm skeptical that this optimization is needed at all.

> oak-upgrade long name filter should consider the path length
> ------------------------------------------------------------
>                 Key: OAK-7335
>                 URL: https://issues.apache.org/jira/browse/OAK-7335
>             Project: Jackrabbit Oak
>          Issue Type: Task
>          Components: upgrade
>            Reporter: Tomek Rękawek
>            Assignee: Tomek Rękawek
>            Priority: Major
>              Labels: candidate_oak_1_8
>             Fix For: 1.9.0, 1.10, 1.6.11
> The NameFilteringNodeState is used to prevent exceptions related to migrating 
> nodes with too long names, which are supported on SegmentMK but not on the 
> DocMK.
> Right now it uses a hard-coded limit of 150 bytes for a name. However, this 
> limit should be only applied if the path length is longer than 350 bytes, so 
> the oak-upgrade should check the path length as well, so it's not too strict.
> See 
> [Utils.java|https://github.com/apache/jackrabbit-oak/blob/5a0fbc54517c8c2af214b2f6f469856016b660d9/oak-store-document/src/main/java/org/apache/jackrabbit/oak/plugins/document/util/Utils.java#L74-L91]
>  for more details.

This message was sent by Atlassian JIRA

Reply via email to