Subject: Re: Progress meter for fsck, revisited
To: Bill Studenmund <wrstuden@NetBSD.org>
From: Nathan J. Williams <nathanw@wasabisystems.com>
List: tech-userlevel
Date: 01/14/2004 00:25:09
Bill Studenmund <wrstuden@NetBSD.org> writes:
> Well, I disagree. I think it's a cool thing to have in the tree, and it
> can benefit a number of users. Also, it is an option, after all.
"A number". How many? Enabling it hurts performance whenever you have
more than one disk.
And again, the fact that it is an option is not justification. I could
add my undergraduate transcript to a kernel file, inside #ifdef
NATHANW_TRANSCRIPT. It's an option. Does that make it OK?
> And, it can serve as a building block/example for a parallel progress
> meter implementation in the future.
The whole problem is that it isn't parallel at all. If it were a meter
of the progresss of parallel processes, I'd be just fine with it.
- Nathan