[
https://issues.apache.org/jira/browse/NIFI-10379?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tamas Palfy updated NIFI-10379:
-------------------------------
Description:
*FechGoogleDrive* can read input from flowfile contents as records.
However the following cases are not handled correctly:
# There is no property on the processor to configure what field in the records
are to be used as file identifiers (based on which fetch is executed). Instead
it is a hardcoded string value.
*Edit:* Upon further consideration the record-based approach poses technical
challenges if we wanted to be properly customizable.
Instead we will consider this to mainly support the output of the
ListGoogleDrive processor. Still can be used from differenc sources but the
expected schema is pre-defined.
# It is possible for the processor to fetch successfully some files but later
experience a general error which sends the incoming flowfile to an error
relationship. This is a mixed result (some records succeed but then _all
records_ fail) and should not happen. Either all records should be processed
(resulting one flowfile for each sent to a success or an error relationship) or
none of them (resulting only the incoming flowfile being sent to another error
relationship).
# _session.xxx(flowFile)_ calls are used as if those were _viod_ (and operated
on the incoming flowFile). Should use the returned flowfile instead.
was:
*FechGoogleDrive* can read input from flowfile contents as records.
However the following cases are not handled correctly:
# There is no property on the processor to configure what field in the records
are to be used as file identifiers (based on which fetch is executed). Instead
it is a hardcoded string value.
# It is possible for the processor to fetch successfully some files but later
experience a general error which sends the incoming flowfile to an error
relationship. This is a mixed result (some records succeed but then _all
records_ fail) and should not happen. Either all records should be processed
(resulting one flowfile for each sent to a success or an error relationship) or
none of them (resulting only the incoming flowfile being sent to another error
relationship).
# _session.xxx(flowFile)_ calls are used as if those were _viod_ (and operated
on the incoming flowFile). Should use the returned flowfile instead.
> FechGoogleDrive record-based input handling improvement
> -------------------------------------------------------
>
> Key: NIFI-10379
> URL: https://issues.apache.org/jira/browse/NIFI-10379
> Project: Apache NiFi
> Issue Type: Bug
> Reporter: Tamas Palfy
> Assignee: Tamas Palfy
> Priority: Major
>
> *FechGoogleDrive* can read input from flowfile contents as records.
> However the following cases are not handled correctly:
> # There is no property on the processor to configure what field in the
> records are to be used as file identifiers (based on which fetch is
> executed). Instead it is a hardcoded string value.
> *Edit:* Upon further consideration the record-based approach poses technical
> challenges if we wanted to be properly customizable.
> Instead we will consider this to mainly support the output of the
> ListGoogleDrive processor. Still can be used from differenc sources but the
> expected schema is pre-defined.
> # It is possible for the processor to fetch successfully some files but
> later experience a general error which sends the incoming flowfile to an
> error relationship. This is a mixed result (some records succeed but then
> _all records_ fail) and should not happen. Either all records should be
> processed (resulting one flowfile for each sent to a success or an error
> relationship) or none of them (resulting only the incoming flowfile being
> sent to another error relationship).
> # _session.xxx(flowFile)_ calls are used as if those were _viod_ (and
> operated on the incoming flowFile). Should use the returned flowfile instead.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
