On Fri, Apr 27, 2012 at 12:16 PM, David Miller <davem@xxxxxxxxxxxxx> wrote:
>
> BTW, I want to clarify that my position is that userland might have
> been the best place to handle this rather than the kernel.
It definitely is not.
You are ignoring the fact that the user-land workaround resulted in an
independent bug in *another* user-land project.
Your whole argument was apparently that "that point in user-land is
the only place that knows the details". But that argument is bogus
crap - and we can *see* that it is bogus crap by the fact that there
was *another* user-land place that had this problem, but didn't
realize it.
That's why fixing it in user land was always wrong. There isn't just a
single user of this interface!
Btw, I think I may have an alternate fix. Which is to simply add a
"packetized" mode to pipes.
Linus
--
To unsubscribe from this list: send the line "unsubscribe autofs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
