Hi Steven
On Sun, Nov 28, 2021 at 5:01 PM Steve973 <steve...@gmail.com> wrote: > > Hi, Claus. I have done what I hope was a fairly full tracing through the > code of camel-core, and I can now say that I appreciate your concern about > adding things to core! I also took a closer look at the Saga and Kamelet > components, and that helped to correct a misconception that I had -- > components are not solely for implementing things like transport > implementations into Camel, and that it is (arguably) a more suitable > mechanism for adding anything that does not have a strict and solid reason > for existing in core. So I will implement this EIP as a component, and not > inside of core. I created a markdown file with my notes when I traced > through core. If you, or anyone else, might be interested, I can attach it > to a reply here. > Thanks for your patience and impressive to hear that you dived into the deep source code and found the good, bad, "and ugly" of our huge code base. One "bad" about EIPs is that the model is defined tightly into the same JAR which means that adding new EIPs is harder - or that users can't develop custom EIPs and plugin as they need. But on the other side then everybody of Camel has the same DSL, and therefore its "standard" and familiar. About your new dynamic eip, then did you create a JIRA for it? If not you are welcome to do so. And then attach the notes in the JIRA so its not lost, and we have it tracked for everybody to see, and also easier find in years to come. I am a bit busy this week with Red Hat virtual company event the entire week. And on the same time I must finish some work on health checks. However I should get some "clear timer" later in December. And we need Camel 3.14 LTS released first also. So have some patience, and then we have better clear road for 3.15 onwards to add new EIPs. > Thanks, > Steve > > On Wed, Nov 24, 2021 at 10:20 AM Claus Ibsen <claus.ib...@gmail.com> wrote: > > > Hi > > > > Adding EIPs or changes to the DSL is not a Camel end user task, and is > > not documented thoroughly. > > Adding new EIPs happens more rarely and there is maybe a little > > hesitation to add because when its in the DSL it stays for a long > > time. > > Also the DSL can become bloated - there are some mistakes from Camel > > v1 and v2 that we have today. > > > > That said to add a new EIP there are 2 ways > > > > - The EIP is implemented fully in the core, so all its code are in core. > > - The EIP is implemeted in a component and are located in camel-xxx in > > the components folder. Examples are circuit breakers, saga, kamelet > > and maybe others I forgot. > > > > So first you need to find out if its the former or the latter. However > > for both cases, the the DSL model are in the core and there is a fair > > bit of code to write to add a new model. > > Therefore its a good idea to find a simple EIP and see where its used > > in the code - can be ConvertBodyDefinition (are in core). Or if its > > outside then maybe KameletDefinition which is the latest EIP added. > > > > The reifer is the "guy" that sits between the model (DSL) and the > > processor (implementation), so all 3 are separated in their own JARs. > > > > And on top of that then there is also the new YAML DSL, so adding a > > new EIP can require some coding there too, but that can be done > > afterwards. > > At first is to get the regular Java and XML DSLs to work. The XML DSL > > is source generated with a maven plugin, and it can "have weird > > problems" if the DSL model is wrong. > > So if you have a very complex EIP then it can be tricky sometimes. > > > > > > > > On Mon, Nov 22, 2021 at 10:16 PM Steve973 <steve...@gmail.com> wrote: > > > > > > Hi, all. I am working on a new version of the Dynamic Router processor > > > that implements the pattern spec much more closely, and I have the > > initial > > > implementation complete, but I want to test it (among other ways) by > > using > > > it in the DSL. I haven't seen any documentation on how to do this, or > > any > > > tutorials. Have I missed something? Could anyone give some kind of > > > overview? I could probably use a different processor and trace up > > through > > > the classes to see what all is involved, but I don't want to miss > > anything, > > > and this would be great documentation to have. If it would be helpful > > for > > > other contributors, I would not mind putting it together for inclusion in > > > the overall documentation. > > > > > > Thanks, > > > Steve > > > > > > > > -- > > Claus Ibsen > > ----------------- > > http://davsclaus.com @davsclaus > > Camel in Action 2: https://www.manning.com/ibsen2 > > -- Claus Ibsen ----------------- http://davsclaus.com @davsclaus Camel in Action 2: https://www.manning.com/ibsen2
