Punt on Spark and Dropwizard

Their dependency trees are large and messy. They include artifacts
with classes that overlap different artifacts elsewhere in the
ecosystem. It is not feasible to address all the problems without
fixing upstream dependency issues.

Author: ctrueden

pom.xml

@@ -206,11 +206,9 @@
 		* ome:jxrlib-all + javax.xml.bind:jaxb-api or jakarta.xml.bind:jakarta.xml.bind-api
 		* ome:jxrlib-all + org.glassfish.jaxb:jaxb-runtime
 		* ome:jxrlib-all + org.glassfish.jaxb:txw2
-		* com.codahale.metrics:metrics-logback + io.dropwizard.metrics:metrics-logback
-		* com.codahale.metrics:metrics-graphite + io.dropwizard.metrics:metrics-graphite
 		* com.github.jnr:jnr-ffi + org.jruby:jruby-core
 		-->
-		<allowedDuplicateClasses>META-INF/versions/9/javax.xml.bind.ModuleUtil,com.codahale.metrics.graphite.*,com.codahale.metrics.logback.InstrumentedAppender,com.sun.istack.*,com.sun.xml.bind.*,com.sun.xml.txw2.*,javax.xml.bind.*,jnr.ffi.*,org.apache.hadoop.yarn.*.package-info,org.apache.spark.unused.UnusedStubClass,org.hibernate.stat.ConcurrentStatisticsImpl,org.junit.runner.Runner</allowedDuplicateClasses>
+		<allowedDuplicateClasses>META-INF/versions/9/javax.xml.bind.ModuleUtil,com.com.sun.istack.*,com.sun.xml.bind.*,com.sun.xml.txw2.*,javax.xml.bind.*,jnr.ffi.*,org.apache.hadoop.yarn.*.package-info,org.apache.spark.unused.UnusedStubClass,org.hibernate.stat.ConcurrentStatisticsImpl,org.junit.runner.Runner</allowedDuplicateClasses>
 
 		<!--
 		NB: We override these properties to make the enforcer happy.
@@ -234,6 +232,19 @@
 		If you need to override the version of a particular component, then simply
 		override that component's version property in your POM's <properties>
 		section, rather than hardcoding them.
+
+		WARNING: While most artifacts in this BOM can be successfully combined,
+		there are a handful of exceptions requiring some downstream effort:
+
+		- net.imagej:imagej-server depends on Dropwizard, which ships its own
+		  metrics-graphite and metrics-logback libraries, among other issues.
+		- org.apache.spark:spark-core_2.11 has a large number of dependencies,
+		  some of which are incompatible with other artifact trees in this BOM;
+		  e.g., Spark depends on kryo-shaded while Bio-Formats depends on kryo.
+
+		If you make use of any of the above artifacts, you may need to add
+		dependency exclusions, ignore additional duplicate class patterns,
+		and/or manually tune your runtime classpath.
 		-->
 
 		<!-- SciJava - https://github.com/scijava -->