[jira] [Commented] (NIFI-10236) ByteArrayContentRepository in core framework

["Matthieu RÉ (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NIFI-10236?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17601383#comment-17601383 ] 

Matthieu RÉ commented on NIFI-10236:
------------------------------------

Hello David, hello Mark,
Thank you very much for your answers ! I needed the whole summer to compute your response and to produce something that passes my tests, sorry for the long time without answer. Now that I have a version that works we can maybe discuss it a little bit more.

My main point is that I struggled to have a wrapped version containing the length for it to be used in the ContentClaimCache, because I wasn't sure about where to set this wrapper for any other component to be able to call the ContentClaim.getLength() or "ContentClaimSomething".getLength(), without this cache available. It seemed more natural to me (and to be honest I didn't managed to make a proper wrapped version work correctly) to store this "length" information directly in the ContentClaim. So this version (https://github.com/apache/nifi/pull/6372) proposes to have the "length", its getter and setter directly in the ContentClaim interface. But then I am not sure about its name, since this "length" is not the actual length taken by the claim in the repository (on my understanding it is more related to an offset ?), and is assigned by another object (the ContentClaimCache for instance)... Should I try another name ? Or maybe do you see a better implementation with a wrapper for the StandardContentClaimWriteCache ?

Thank you for you advices

> ByteArrayContentRepository in core framework
> --------------------------------------------
>
>                 Key: NIFI-10236
>                 URL: https://issues.apache.org/jira/browse/NIFI-10236
>             Project: Apache NiFi
>          Issue Type: New Feature
>          Components: Core Framework
>    Affects Versions: 1.17.0
>            Reporter: Matthieu RÉ
>            Assignee: Matthieu RÉ
>            Priority: Major
>          Time Spent: 1h 20m
>  Remaining Estimate: 0h
>
> As discussed in NIFI-8760 and by mail, the use case of using an in-memory based implementation of the ContentRepository is not completed after 1.17.0 and the removal of the old VolatileContentRepository (NIFI-9823) when the ExecuteStateless is not a solution.
> As proposed by mail, the ByteArrayContentRepository could be duplicated from the nifi-stateless framework to the core one, but maybe with more functionalities, for instance the follow-up of the space taken in memory, and the limitation of its total usage.
> Don't hesitate to comment, add or discuss requirements or the implementation I will propose.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)