| Summary: | SafariWebsocketDraft0Test failure during build | ||
|---|---|---|---|
| Product: | [RT] Jetty | Reporter: | Jan Bartel <janb> |
| Component: | websocket | Assignee: | Simone Bordet <simone.bordet> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | gregw, jetty-inbox, tbecker |
| Version: | 8.1.0 | ||
| Target Milestone: | 7.5.x | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| Whiteboard: | |||
|
Description
Jan Bartel
Jan, is this reproducible for you? I've tried to reproduce this by doing a whole build a couple of times without success and then I've run the individual tests for a while in a loop like this: while [ $? -eq "0" ] ; do mvn -Dtest=SafariWebsocketDraft0Test test ; done It never fails for me. (macos, jdk1.6.0_29, mvn 3.0.3). If it's an intermittent failure, then I will dig a bit more into the test and it's timings. The test was incorrectly using a BufferedReader *and* the input stream to read the bytes. It was possible that the hixie bytes were already read by the BufferedReader when the test attempted to read them from the input stream, causing the read to timeout. note that this fix was applied to both jetty-7 and jetty-8. We should still be avoiding direct jetty-8 changes for anything but servlet 3.0 stuff so changes from 7 can be merged to 8. |