On 9/26/07, Saladin Sharif <[EMAIL PROTECTED]> wrote: > triggered a JVM thread dump on the hanging Ant > process. I have copied and pasted the thread dump at > the end of this e-mail message.
Nothing that looks unusual there. The cvs command was forked, and the stream pumper on another thread is reading the output of the command. In the past (months? years?), some people complained cvstagdiff was taking an inordinate amount of time compared to the same command on the command line. You haven't said anything on how long the command takes on the CLI, and whether this time changes or not with <exec>. You could try testing on a smaller repo or inside a subset of your repo. If someone the command slows down, how long have you waited before asserting it hangs? Doing the test on a very small repo might allow it to complete in a "reasonnable" time. Also, try thread dumps at different times, to see if you can get a different stack-trace. > > 2) Provide relevant information: version of CVS > > client? version of CVS > > server? JVM? Ant version? OS? etc... > > * CVS client version: > > Concurrent Versions System (CVS) 1.11.22 (client) I'd try CVSNT. I've always used it on Windows, and used to do cvstagdiff with it (a long time ago). I get it from TortoiseCVS, rather than the CVSNT stand-alone install. > Thanks for the suggestions, I'm running out of ideas. If nothing works, modify cvstagdiff to work off a cvs diff file output separately by a different command, and combine it with an <exec>. There's a remote possibility that your server outputs something that throws off the parsing code of cvstagdiff, but the thread dump does not point to such an issue. --DD --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
