Mark Hiner

It is possible that we could solve this as a higher level... I assume it is still due to the polling thread set up by SCIFIO... but I couldn't come...

You can just provide the license in-repo and reference it via the maven config.. see [scifio-tutorials](https://github.com/scifio/scifio-tutorials)

This needs to be robustly tested - e.g. if running from a JRE can the compiler in a JDK be located? Does using a different java location (like internet plugins)...

First bit of progress: I have a [minimaven branch](https://github.com/scijava/minimaven/pull/15). Using it, I was able to successfully compile a .java file from the script editor after removing: 1. tools.jar 2. javac.jar...

note that compilation still works _with_ these jars as well

>so at the moment I believe the native launcher is rolled back to the pre-snazzy version that doesn't work so well in this regard. @ctrueden that is correct. @papalagirauscher there...

As we update the build system the scope of this issue will move beyond ImageJ.cfg (which should go away). We want the ability to designate a system and user jar/plugin...

For after https://github.com/scijava/scijava-common/issues/96

@karlduderstadt @ctrueden it looks like the problem is that the SciJava IO plugin was always completely claiming the IO process ([returning a non-null value even if no handler was found](https://github.com/imagej/imagej-legacy/blob/d2a00f7f8779a717eb632a7d59e1fe047a191a4a/src/main/java/net/imagej/legacy/plugin/DefaultLegacyOpener.java#L132))....

@karlduderstadt I'm sorry, I totally misunderstood the problem. From what I can tell, I agree with you about the location of where to slot in the `uiService.show` call, and [opened...