[ 
https://issues.apache.org/jira/browse/NIFI-13997?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17907145#comment-17907145
 ] 

ASF subversion and git services commented on NIFI-13997:
--------------------------------------------------------

Commit 116f2f728bd13e659f0725d4f238537b4ff40a89 in nifi's branch 
refs/heads/main from Ferenc Gerlits
[ https://gitbox.apache.org/repos/asf?p=nifi.git;h=116f2f728b ]

NIFI-13997 Limit Python Processor loading to requested Processor Class (#9568)

Signed-off-by: David Handermann <[email protected]>

> When loading a Python processor, do not load other processor code
> -----------------------------------------------------------------
>
>                 Key: NIFI-13997
>                 URL: https://issues.apache.org/jira/browse/NIFI-13997
>             Project: Apache NiFi
>          Issue Type: Improvement
>            Reporter: Ferenc Gerlits
>            Assignee: Ferenc Gerlits
>            Priority: Minor
>          Time Spent: 1h 10m
>  Remaining Estimate: 0h
>
> When NiFi loads the code for a Python processor contained in a package, it 
> loads all Python files in that package (directory). This means that the 
> dependencies of all processors in the package must be identical.
> In some cases, the user may want to have the ProcA and ProcB processors in 
> the same package to have some common dependencies dep_common (listed in the 
> {{requirements.txt}} file) plus each processor to have some extra 
> dependencies, say dep_a and dep_b (listed in 
> {{ProcessorDetails.dependencies}} in each processor module). When loading 
> ProcA, we currently only install the dep_common + dep_a dependencies, but 
> then try to load both ProcA and ProcB, which will fail due to the missing 
> dep_b dependencies.
> To solve this problem, when we load a processor from a package, we should not 
> load any other processors in the same package. We should load all 
> non-processor helper files plus the single processor requested by the user.



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

Reply via email to