[Twisted-Python] Re: [Twisted-commits] r16582 - Remove util.wait and all uses.

[Andrew Bennetts]

On Mon, Apr 10, 2006 at 01:10:30AM -0400, Jean-Paul Calderone wrote:
[...]
> >--- trunk/twisted/protocols/ftp.py	(original)
> >+++ trunk/twisted/protocols/ftp.py	Sun Apr  9 21:21:33 2006
> >@@ -660,7 +660,7 @@
> >            if cmd == 'PASS':
> >                return self.ftp_PASS(*params)
> >            else:
> >-                return BAD_CMD_SEQ
> >+                return BAD_CMD_SEQ, "PASS required after USER"
> >
> >        elif self.state == self.AUTHED:
> >            method = getattr(self, "ftp_" + cmd, None)
> >@@ -672,7 +672,7 @@
> >            if cmd == 'RNTO':
> >                return self.ftp_RNTO(*params)
> >            else:
> >-                return BAD_CMD_SEQ, "Blah"
> >+                return BAD_CMD_SEQ, "RNTO required after RNFR"
> >
> >
> >    def ftp_USER(self, username):
> >
> 
> Hmmm.  I don't see any changes related to wait() in these hunks. ;)

This was my bad (I did the ftp parts of this branch) -- I bumped into the bug in
the first hunk while updating test_ftp, and improved the nearby "Blah" while I
was at it :)

[...]
> 
> While it doesn't appear that this change *introduced* a new race condition, 
> it does seem as though it has jostled things so as to cause an existing one 
> to start going the wrong way.
> 
[...]
> Please revert the branch or fix the race post-haste.  If someone would take 
> a look at the failing tcp half-close test on that reactor, I would 
> appreciate it, otherwise I'll try to get to it sometime in the next couple 
> days.

Jonathan and I both came up with essentially the same fix for this.  He's just
committed mine to the trunk in r16583.

Thanks for spotting this.

No idea about the half-close test, though...

-Andrew.