I was looking into https://issues.apache.org/jira/projects/DOXIATOOLS/issues/DOXIATOOLS-68 and https://issues.apache.org/jira/projects/DOXIATOOLS/issues/DOXIATOOLS-67 when Modello got in my way. So I tried a little experiment to see how hard it would be to rip Modello out of a project. Result: not very hard. Just copy the generated sources into the src tree, remove the .mdo files and plugins, and add some license comments to make Rat happy.
Why do this? Several reasons: 1. Eclipse can't build Modello based projects. Yes there's probably some magic incantation and knobs to twist in the settings which can make it work if you look at it sideways, but why go through that hassle every time you want to set up a new project. 2. Modello is an extra thing for developers to learn and understand before they can contribute to a project. Java is well understood. Modello isn't. It's only really used by Maven. Let Java developers write and edit Java code. It's not as if writing value classes with equals methods is hard, and IDEs can autogenerate a lot of what Modello gives us. 3. Modello is one more tool that can break when building a project because of a new JDK, a hosting site going dark, etc. Tools of this nature tend to break more frequently than pure Java code or Maven itself. The build is less fragile without it. What do folks think about slowly moving away from Modello where feasible to simplify the build? Does anyone find Modello a net positive, especially in longterm maintenance, not just in initial code generation? -- Elliotte Rusty Harold [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
