[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

RE: cygwin/windows or test logfiles



> -----Original Message-----
> From: owner-or-dev@freehaven.net [mailto:owner-or-dev@freehaven.net] On
> Behalf Of Roger Dingledine
> Sent: Thursday, June 12, 2003 5:38 AM
> To: or-dev@freehaven.net
> Subject: Re: cygwin/windows or test logfiles
> 
> On Wed, Jun 11, 2003 at 10:33:45AM -0400, Bruce Montrose wrote:
> > TEST DESCRIPTION:
> >
> > 1. I first started up or1 and or2 which are known to work.
> > 2. I generated traffic through or1 and or2 with my browser to verify
> >    that or1 and or2 were working properly.
> > 3. I then started up or4 which is the cygwin version.
> > 4. I attempted to generate more traffic through the onion routers.
> > 5. I noticed quite a lot of activity on all three onion routers, but
> >    ultimately the cygwin onion router eventually hung.
> 
>  From the log files, it looks like you didn't try doing anything after it
> hung. So it could just be that it read the whole file off the webserver,
> and then it didn't recognize eof? I'm curious whether you could have
> sent another transaction through it at that point.

I'll check on the eof issue. I tried sending another transaction after the
cygwin OR hung unsuccessfully and with no activity in any of the log files.

> 
> > 6. I then killed the cygwin onion router and the traffic that I had
> >    queued up went through the other two onion routers at that time.
> 
> This doesn't seem consistent with the log files. What do you mean other
> traffic that you had queued up?

What I meant to say is that the transaction initiated by my browser that
caused the cygwin onion router to hang and the browser to be perpetually
waiting for the transaction to complete, eventually did complete immediately
after I killed the cygwin onion router (ie. The browser displayed the
results that it received from the onion routers).

> 
> But the overall result of this convinces me that it's going to be
> considerable (meaning more than an afternoon, who knows how much more)
> work to get this thing running on Windows. The reason for this is that
> Windows doesn't do things in a Posix way, and I don't know how much they
> deviate. I'm tempted to put it off until we have users, since then at
> least we would have people to tell us when we get it right; and if we're
> lucky some of them would be Windows savvy and could give us a hand.
> 
> Bruce -- you don't happen to have experience with network I/O on Windows,
> by any chance? :)

Yeah, but it's nothing something I'm proud of :) I'll take a look and see
what I can find.

> 
> Thanks,
> --Roger