Proposed Change - RFC - net-tools to iproute2

Bryan Kadzban bryan at kadzban.is-a-geek.net
Sat Jun 26 19:52:27 PDT 2004


Jim Gifford wrote:
> 2 - Using coreutils hostname (Need to remove hostname from the 
> suppress patch)

Does "hostname -f" work properly with coreutils' hostname yet, or does
it still set your hostname to -f like it did when it was part of the
sh-utils package?

I wonder if it would be easy to use the hostname from net-tools anyway?
Or submit a patch to the iproute2 people that moves it over?  (I haven't
tried any of this; the iproute2 people might not even want to supply a
hostname binary in the package...)

How many bugs are in net-tools that aren't being fixed because it hasn't
been changed, btw?  Or is the lack of changes due to it being mature
enough to not need updates?  (Granted, this is probably unlikely, but I
never like it when people claim a package is "unmaintained" simply
because it hasn't been updated.  There are only so many things that a
sane hostname/arp/whatever program can do...)

You state that "it does affect performance".  TBH, you're going to have
to give more details to convince me (for whatever convincing me is
worth).  What have you seen?  What is faster?  (I assume you mean it
does actually speed stuff up; you haven't said that for sure though ;-))
How much faster?

What are the differences between netstat and nstat (or is nstat
something completely different anyway, and if so, what replaces
netstat)?  Replacing netstat with nstat and ifconfig with ip <something>
is also going to throw a bunch of people off stride, I think.  netstat
and ifconfig (with those names) are pretty standard in Linux, as far as
I can tell anyway.  And no, this isn't a "well these distros do it"
argument -- I'm saying if we're too different, all we'll do is confuse
people unnecessarily.  Now, having both installed at the same time
wouldn't be bad (to me anyway), but I don't think a lot of other people
agree with me on that one.

All that said, it looks like there's some good new functionality to be
had in the iproute2 package (tc, for example).  I don't suppose it would
make sense to let the reader choose, would it.  No, probably not (that
would increase support traffic for network issues, and probably cause
issues with BLFS also), rats.



More information about the lfs-dev mailing list