[hibernate-dev] Build and AnnotationProcessors

[Steve Ebersole]

For the Eclipse side its actually pretty simple to add this capability 
of generating all sources when the project is generated since the 
recommendation seems to be to do this from the Gradle side (`gradlew 
eclipse`).  We'd just need to add a doFirst action to call the 
generateSources task.  There is already this[1], which may or may not do 
that (dunno).

For IDEA, especially now that its Gradle integration is getting so good, 
thats no longer the case.  I actually now always use the IDEA "import 
gradle project" functionality.


[1] eclipseClasspath.dependsOn( "generateSources" )


On 04/17/2013 01:28 PM, Eric Dalquist wrote:
> Just to chime in on the IDE use. While hibernate is on the significantly
> complex side of things for OSS projects go most projects I deal with
> that are using Maven or Gradle I do just checkout in my IDE and expect
> to work.
>
> For 99% of maven projects this is the case, I checkout/clone the
> project, import it as a maven project in Eclipse and I'm done. All code
> and resources required to build are correctly setup on the classpath and
> I'm good to go.
>
> For 99% of gradle project I do something similar but before the import I
> just run the gradle eclipse command on the CLI.
>
> Unless I'm going to be doing active development or patching on a project
> I don't care about a CLI build, I care about a functional IDE experience
> so I can run unit tests and debug code to verify some behavior/problem/bug.
>
> I don't have much to say specifically on the annotation processors
> stuff. Right now all the jasig projects use a combination of the
> maven-processor-plugin and the build-helper-maven-plugin to get the
> meta-model built and added to the classpath. Honestly I don't care how
> that happens as long as I can end up with the same final result of maven
> knowing about a generated-sources directory and code getting placed
> there during the generate-sources build phase.
>
> https://github.com/UW-Madison-DoIT/BlackboardVCPortlet/blob/uw-master/blackboardvc-portlet-webapp/pom.xml#L284
>
> -Eric
>
> On 04/17/2013 11:58 AM, Steve Ebersole wrote:
>> On Wed 17 Apr 2013 11:44:18 AM CDT, Hardy Ferentschik wrote:
>>> On 17 Jan 2013, at 6:30 PM, Sanne Grinovero <sanne at hibernate.org> wrote:
>>>
>>>> AFAIK the reason we originally had a separate phase for annotation
>>>> processors was to workaround the following javac bug:
>>>> http://bugs.sun.com/view_bug.do?bug_id=6512707
>>>   From a Maven perspective (not an ORM issue longer) there is/was also
>>> the issue of configuring the annotation processors in the compiler plugin.
>>> There are several unresolved Jira issues for that, but according to David
>>> they are resolved, even though I cannot see any reference of that in the
>>> Maven issue tracker.
>> Right, would not affect ORM anyway.
>>
>>
>>>> On command line users.. I'd agree with Hardy that we likely all prefer
>>>> building from the command line rather than from the IDE, but I don't
>>>> expect that to be the majority of users.
>>> So what do you do when you are interested in some project and want to check it out?
>>> Get the sources and load it directly in your IDE. IMO that is stupid. I first build at least
>>> once from the command line and have a look at the generated directories. Does the
>>> build work? Can I make sense off things w/o knowing much about the code?
>>> Only then I would start using an IDE and I expect that a developer can set up his
>>> IDE of choice.
>> So when someone does something different than you they are stupid...
>> nice :)
>>
>> _______________________________________________
>> hibernate-dev mailing list
>> hibernate-dev at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev