On 08/11/2013 11:01, Konstantin Kolinko wrote: > 2013/11/7 <ma...@apache.org>: >> Author: markt >> Date: Thu Nov 7 19:24:17 2013 >> New Revision: 1539771 >> >> URL: http://svn.apache.org/r1539771 >> Log: >> Add a workaround for a Jira bug (to be removed later) as Jira is being used >> for testing >> > > Ordering of jars is important. The ones in <PreResources> should be > listed first.
The ordering of the groups is important but - and this was the Jira bug - apps should not be depending on a particular priority order for JARs in WEB-INF/lib. Retaining the order in listResources is probably the simplest way of implementing this. > (VirtualWebappLoader of Tomcat 7 can inject jars to be searched before > ones in WEB-INF/lib or after them, depending on configuration. One > should be able to configure the same in Tomcat 8). Agreed. > The same s/HashSet/LinkedHashSet/ is likely true for the > implementation of #listWebAppPaths() as well, though I am not sure > about an use case there. I don't think so as it is only the paths in String form that are returned. The required ordering would be enforced when one of those paths was looked up. > One more note on #listWebAppPaths(), looking at a method that uses it: > > Javadoc of method ApplicationContext.getResourcePaths(String) says > that the returned set is immutable. This statement is not true in > Tomcat 8, as far as I see it returns the created mutable HashSet as > is, > > Actually, I do not see such immutability requirement in Javadoc of > ServletContext (neither in ours nor in the one at Oracle site), > http://docs.oracle.com/javaee/7/api/javax/servlet/ServletContext.html I fixed the Javadoc. Thanks for the review. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
