Greg Wilkins

@sbordet Currently the problematic API is: ```java protected DelayedProcess newDelayedProcess(boolean contentExpected, String contentType, MimeTypes.Type mimeType, Handler handler, Request request, Response response, Callback callback) ``` So first question: Should this handle...

@sbordet I'm bumping this to 12.1.0 and think we should consider re-introducing the built in HTTP/1 delayed dispatch (on by default) and making this handler less general purpose.

@sbordet @lorban @lachlan-roberts I'm going to reintroduce delayed dispatch to HttpConnection. Can you guys look at doing the same for h2 and h3. I think the DelayedHandler should then only...

@sbordet I just noticed that all the EE10 DosFilter tests are disabled. So can we make this issue a priority for the next release cycle.

@joakime The problem is that the current code will return a violation description OR report it to a listener. The proposed code will report a violation to the listener and...

@joakime I think something like the following would be a better implementation: ```java public static String checkUriCompliance(UriCompliance compliance, HttpURI uri, ComplianceViolation.Listener listener) { if (uri.hasViolations()) { StringBuilder violations = null;...

Oh! are we only reporting violations to listeners if they are allows by a compliance mode? Maybe we should always report them to a listener, but only throw if they...

@joakime would the following be better: ```java public static String checkUriCompliance(UriCompliance compliance, HttpURI uri, ComplianceViolation.Listener listener) { if (uri.hasViolations()) { StringBuilder violations = null; for (UriCompliance.Violation violation : UriCompliance.Violation.values()) {...

Let's fix these tests in 12.1.0 and stop kicking the can along the road.

The servlet spec requires that we commit the response as soon as precisely the number of bytes written matches any content-length set. So we do need to track bytes written,...