On 18/02/2015 18:56, Mark Thomas wrote: > I'm currently exploring explicitly triggering the onWritePossible() and > onDataAvailable() calls but I have some more work to do on this as the > work-arounds needs to be removed to avoid double registration etc.
A short progress report. It has been impossible to separate fixing the onWritePossible() and onDataAvailable() calls from: - supporting half-close for upgraded connections and the correct timing of HttpUpgradeHandler.destroy() - ServletInputStream.read() should return -1 at end of stream - multiple attempts to close the same socket I'd fix one thing only for the fix to expose a different bug. I also found and fixed a potential infinite loop in SecureNio2Channel. I'm currently running the unit tests locally to confirm that the current set of fixes doesn't trigger any further regressions. Once those are passing on at least one OS, I'll commit them and let the CI system do its work. I'll try and separate each of the fixes into a separate commit but there is quite a lot of overlap. While the tests are running I'll see if I can put together a git branch with a series of smaller commits rather than one big one. I haven't yet thought about how/if this will be back-ported to 8.0.x. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org