Hi See also https://issues.apache.org/jira/browse/CAMEL-26285
On Sat, Oct 24, 2026 at 10:25 PM Claus Ibsen <[email protected]> wrote: > Good point, and it's already partly there: each body type can carry a > javaType with the fully qualified class name, and when the type depends on > an option, such as operation=putObject, each option value can have its own > javaType. Around 160 component sides have a single type, and there are > around 870 per-option types. > > Milvus is a good example of what isn't covered yet. The action comes from > the CamelMilvusAction header rather than an endpoint option, so today the > SDK classes (InsertParam, SearchSimpleParam, MutationResult, SearchResponse > …) are only named in the description. I'll look at letting the per-option > types also key on a header, so Milvus and similar components can list the > fully qualified class name for each action. > > > > On Sat, Oct 24, 2026 at 6:21 PM Federico Mariani < > [email protected]> wrote: > >> Hi Claus, >> >> That's really nice, what if it is a java object? would it be worth to have >> the full qualified name? IIRC components like milvus interact with Milvus >> specific objects both as input to the component and output. >> >> Il giorno sab 24 ott 2026 alle ore 15:50 Claus Ibsen < >> [email protected]> >> ha scritto: >> >> > Hi >> > >> > I am working my way from A..Z and have soon all the components as a >> bunch >> > of PRs. >> > After that its the EIPs and then we have most of it. >> > >> > On top of that the agents help find many small bugs so expected 200+ >> being >> > reported as well. >> > >> > >> > >> > On Thu, Oct 22, 2026 at 9:59 AM Claus Ibsen <[email protected]> >> wrote: >> > >> > > Hi >> > > >> > > For Camel 4.24 we are working on the question that tooling, AI and >> people >> > > ask all the time when writing the next step of a route: what is the >> body >> > > here? >> > > A String, a byte[], a stream, a Map, a GenericFile? (CAMEL-24844) >> > > >> > > It grew out of the local model benchmark, where some of it is already >> > part >> > > of the runs: >> > > >> > >> https://camel.apache.org/blog/2026/10/camel-local-model-benchmark-round-4/ >> > > >> > > At runtime this is already in place: an error about the body says what >> > the >> > > body is and which step made it so (CAMEL-25749), and the message >> history >> > > records the body type at each step (CAMEL-25750), shown by >> > > "camel cmd route-structure", the dev console and the MCP tools. >> > > >> > > Now we are adding the static side (CAMEL-24961, first asked for in >> > > CAMEL-330 >> > > back in 2008): a new @BodyTypes annotation declares the body each >> > > component, >> > > data format and language takes and gives. It goes into the catalog, so >> > the >> > > docs, "camel catalog doc", the MCP tools and the YAML validator know >> it >> > > too. >> > > >> > > This comes as a series of PRs, group by group across the components. >> The >> > > first >> > > ones are open: #28241, #28242, #28246, #28247, #28250 and #28251. >> > > >> > > So expect many more PRs on the way to cover all of the components, and >> > > what else. >> > > >> > > >> > > >> > > -- >> > > Claus Ibsen >> > > >> > >> > >> > -- >> > Claus Ibsen >> > >> > > > -- > Claus Ibsen > -- Claus Ibsen
