[
https://issues.apache.org/jira/browse/CAMEL-26874?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Claus Ibsen resolved CAMEL-26874.
---------------------------------
Resolution: Fixed
Fixed by https://github.com/apache/camel/pull/29700
> camel-jbang - AI tools: the refusal of two EIPs in one step gives the line to
> write
> -----------------------------------------------------------------------------------
>
> Key: CAMEL-26874
> URL: https://issues.apache.org/jira/browse/CAMEL-26874
> Project: Camel
> Issue Type: Improvement
> Components: camel-jbang, camel-yaml-dsl
> Reporter: Claus Ibsen
> Assignee: Claus Ibsen
> Priority: Minor
> Fix For: 4.24.0
>
>
> Follow-up from the local-model benchmark (qwen3.6, r32) and CAMEL-26758 /
> CAMEL-26825. In r32 the model sent the same refused content 3 or more times
> in 9 step runs of the new set. In 5 of them the refusal was "must have at
> most 1 properties": one list item of steps held two EIPs, as in
> {code}
> - log:
> message: "..."
> setBody:
> expression: ...
> {code}
> The message says "setBody: is another EIP: start it as its own item, -
> setBody:", but it gives no line to write, so the loop breaker of CAMEL-26758
> cannot apply it.
> * The message now ends with the line to write: "Write line 47 as:" and the
> key line with "- " in front of the key, at the column of the dash of the
> item. The lines under the key stay as they are. A model can copy the line,
> and the write tool applies it on the third identical write.
> * When doCatch: or doFinally: was written at the column of doTry:, the
> validator called it "another EIP", so its advice was to start it as its own
> item. They are options of doTry: and go under it, next to steps:. In r32 one
> step run sent that content 19 times. An option of the EIP is now said to be
> an option, also when its key is the name of an EIP too.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
