Hi Robert

In IntelliJ components are not managed by an OSGi container and so the 
extension of the Eclipse’s Plugin class and the use of OSGi is troublesome.

I managed to create a Stub so that I can use the code from the project but I 
discovered other usages of the OSGi bundles like in the ArtifactLocatorImpl and 
probably more.
I just started to delve into the issues with using the Sling IDE backend code 
and there is probably more. For now I can proceed but later we should adjust to 
a common codebase.

I will create a JIRA ticket to keep track of it tomorrow.

- Andy

> On May 4, 2015, at 3:47 AM, Robert Munteanu <[email protected]> wrote:
> 
> Hi Andy,
> 
> On Tue, 2015-04-28 at 09:02 -0700, Andreas Schaefer Sr. wrote:
>> Hi
>> 
>> Most of the backend code inside the tooling project (api, impl
>> -resources etc) do not depend on Eclipse with the sole exception of 
>> impl-vlt which use the Eclipse Plugin class inside the Activator and 
>> uses the Activator’s getLogger() to obtain the logger inside that 
>> module.
>> 
>> IntelliJ does not use an OSGi container and even though I could stub 
>> the Eclipse Plugin class it does not seem right.
>> 
>> Can we split that module into 2 parts where one only contains the 
>> backend code and the second would contain the Eclipse Plugin service 
>> code and configuration?
>> 
>> The only sticky issue I see is the call inside classes like 
>> VltRepository to obtain the Plugin Logger: 
>> Activator.getDefault().getPluginLogger(). One solution would be to 
>> create another class like PluginLoggerProvider that has a static 
>> method like getInstance() which has a setter and getter method for 
>> the Plugin Logger.
>> 
>> Cheers - Andy
> 
> Yup, mostly likely we need an indirection for obtaining the
> PluginLogger. Can you file a JIRA issue for this? 
> 
> And just to be sure, from your point of view it's fine that the
> Activator exists, it's just the calls to Activator.getDefault() that
> are troublesome, right?
> 
> Thanks,
> 
> Robert
> 

Reply via email to