Subject: Re: wd woes
To: None <current-users@sun-lamp.cs.berkeley.edu>
From: John F. Woods <jfw@ksr.com>
List: current-users
Date: 11/29/1993 10:46:01
> >>Why hasn't this shown up in current yet?  This patch *needs* to be in
> >>current.  I can testify to its worthiness.
> >actually, i think the patch is a crock. it does not solve the problem,
> >it only hides it.
> >oh, and it has been tried by other people. as said, it does NOT solve
> >the problem!
> >in fact, there are three wd bugs. this `fakes out' only one of them.
> >the patch is also exceedingly ugly.
> Well, all these things may be true, but it's also true that my box
> used to lock up regularly stuck on some IDE hard drive access, and
> since building a kernel with this patch, it no longer does so.  This
> makes my NetBSD box exceedingly more useful.  These are the results.
> Now, it may also be true that this just band-aids over the symptoms.
> But, at least it does that much.

For my own personal perspective on this kind of thing:

Band-aid patches usually cause more problems than they solve (easily true,
if it solves one big problem and causes two little ones :-); first, since
they usually don't really fix the problem but just reduce its frequency, they
give a false sense of security to people who are going to become disappointed
eventually; second, by cutting down on the frequency of bug reports, they
cause overworked programmers to forget about the actual problem (:-); third,
when a half-assed patch gets committed to a shared source base, you often
wind up with people developing over the top of that patch, sometimes making
it difficult to commit a genuine repair later.  Band-aid patches don't often
help in the long run, and often make it difficult to get genuine relief.

> If this patch is so ugly, *please*, someone write something better.
> Until then, I at least desire my box to be useable.

Well, you claim the well-publicised patch works for you; if it does, you
can keep a copy of it handy and re-patch wd.c for yourself until wd.c mutates
beyond recognition.  With luck, that point will happen when the genuine fix
is put in :-).

------------------------------------------------------------------------------