Subject: Re: zs driver still broken?
To: Charles M. Hannum <root@ihack.net>
From: Dr. Bill Studenmund <wrstuden@loki.stanford.edu>
List: port-sparc
Date: 02/01/1999 10:30:57
On Mon, 1 Feb 1999, Charles M. Hannum wrote:
> This change is *still* not correct, for reasons I already stated.
Charles, are you helping or hindering?
I did not say that the patch addressed all the issues you raised. I did
not say it's the end of the story.
Your complaint was that we should do a lot of what the status interrupts
do, like start or stop flow based on the new values. You are, in
principle, right.
However all you have done is complain. Complain about any changes.
Complain that you are not being listened to.
Well Charles, the problem is not that I'm NOT listening to you, it's that
I am. Not only am I hearing technical issues, but I'm also hearing a lot
of anger and frustration from you directed at, in this case, me. The
latter has no place in this discussion. I'm trying to keep my emotions
out, but can only do so much. It'd be a lot easier if you wouldn't convey
so many of your emotions in your messages.
Yes, with 1.53 I made a mistake, which you had pointed out a while back.
I'm fessing up. But I think I've fixed that mistake in 1.56.
Yes, 1.56 has outstanding issues about starting or stoping flow control.
But so did 1.52. And so did all the versions from Nov-97 when you last
worked on this code?
I guess all my frustration and anger can be summed up in one question:
Why, in order to regain functionality I'd put into z8530tty.c before you
revamped it, do I have to impliment functionality which you chose not to
impliment at the time, even though you knew it was missing?
Please, don't get me wrong. I'm not complaining that the functionality's
not there or that you didn't put it there. I'm complaining that you're
demanding I add it before I can do anything else.
So to reitterate: Charles's analysis is right, but I assert that 1.56 is
no less broken than 1.52 (before the clock protection changes), no less
broken than the versions Charles worked on in Nov-97, and no less broken
than the current implimentation in src/sys/dev/ic/com.c. As the latter
cases are (were) ok, 1.56 is not a step in the wrong direction.
Take care,
Bill