WildFly 8/Undertow 1 handles URL fragments differently than JBoss EAP 6.2/Apache-Coyote/1.1 ------------------------------------------------------------------------------------------- Key: WFLY-2969 URL: https://issues.jboss.org/browse/WFLY-2969 Project: WildFly Issue Type: Bug Security Level: Public(Everyone can see) Components: Web (Undertow) Affects Versions: 8.0.0.Final Reporter: Michal Babacek Assignee: Stuart Douglas Priority: Minor Fix For: 8.0.1.Final I have a bunch of URL tests, executed against EAP 6.2.0 directly and then proxied through mod_cluster 1.2.6. These all pass. I tried to run the same with {{Wildfly-8.0.1.Final-SNAPSHOT, HEAD:be3a13e6}} with a surprising result of seeing some differences in how are various Request attributes reported from within the servlet: Here we go,{{8080}} is where either EAP or WFLY listens: || ||{{curl 'http://10.16.88.19:8080/clusterbench/requestinfo//?;?=44&test=OK;devil=3&&&&&&&&&&&&&&&&&&&&&&&&&&&&777=666'}}|| |EAP|Parameters map: {test=OK;devil=3, 777=666, ;?=44}| |WFY|Parameters map: {=, 777=666, test=OK;devil=3, ;?=44}| *Note:* {{Parameters map: \{key=value, key=value, ..., ...\}}}, so there is like an empty key and and empty value for WFLY.... This is the [CommonRequestInfoServlet.java|https://github.com/Karm/clusterbench/blob/s...] that produced the output. {code} Map params = request.getParameterMap(); Iterator i = params.keySet().iterator(); while (i.hasNext()) { String key = (String) i.next(); String value = ((String[]) params.get(key))[0]; responseText.append(key); responseText.append("="); responseText.append(value); if (i.hasNext()) responseText.append(", "); } {code} It is, indeed, possible that these mangled URLs are slipping through the cracks of [RFC|http://tools.ietf.org/html/rfc3986] and should be either "fixed" or URL encoded, yet I find the different interpretations disturbing.