[Twisted-Python] Re: [Twisted-commits] r14952 - Impelement spawnProcess for IOCP reactor. Resolves issue1008.
![](https://secure.gravatar.com/avatar/7ed9784cbb1ba1ef75454034b3a8e6a1.jpg?s=120&d=mm&r=g)
On Fri, 28 Oct 2005 10:38:02 -0600, Justin Johnson <justinj@wolfwood.twistedmatrix.com> wrote:
While I am excited this feature is so close, it seems like it isn't quite there yet. Buildbot is now reporting only 442 tests run by the IOCP builder on Win32, and a really nasty looking traceback afterwards. Could we revert this merge? Jean-Paul
![](https://secure.gravatar.com/avatar/b3407ff6ccd34c6e7c7a9fdcfba67a45.jpg?s=120&d=mm&r=g)
On Fri, Oct 28, 2005 at 11:26:22PM -0400, Jean-Paul Calderone wrote:
The bug you notice existed before that merge too. Jonathan helped me diagnose it: see http://twistedmatrix.com/bugs/issue1294. It appears trial is calling reactor.iterate(0) during cleanup, and not expecting that it could raise an exception. Making iocp's doIteration swallow and log that error, like the other reactors do, will probably fix this. Making trial deal with that error better (as apparently is not too hard to do with one of the trial branches) would also fix it. -Andrew.
![](https://secure.gravatar.com/avatar/b3407ff6ccd34c6e7c7a9fdcfba67a45.jpg?s=120&d=mm&r=g)
On Fri, Oct 28, 2005 at 11:26:22PM -0400, Jean-Paul Calderone wrote:
The bug you notice existed before that merge too. Jonathan helped me diagnose it: see http://twistedmatrix.com/bugs/issue1294. It appears trial is calling reactor.iterate(0) during cleanup, and not expecting that it could raise an exception. Making iocp's doIteration swallow and log that error, like the other reactors do, will probably fix this. Making trial deal with that error better (as apparently is not too hard to do with one of the trial branches) would also fix it. -Andrew.
participants (3)
-
Andrew Bennetts
-
Jean-Paul Calderone
-
Justin Johnson