Am 29.01.2014 09:19, schrieb Michal Kubecek:
On Wednesday 29 of January 2014 08:37:47 Ludwig Nussel wrote:
Note that meanwhile it looks like there is development on net-tools again. There is even a git repo: http://sourceforge.net/p/net-tools/code/ci/master/tree/
Did anyone check whether they already fixed some of the concerns or plan to?
No, but AFAIR, most of our patches are upstream. At least the Author told me that. But because we don't really have a maintainer [just on paper] for net-tools, it is as it is... Nobody has time to care about, see e.g.: https://bugzilla.novell.com/show_bug.cgi?id=821077
It would be definitely possible to address some of them but some are intrinsic and cannot be fixed without breaking backward compatibility which would be IMHO even worse than throwing out the utilities.
Yes, that's the problem with it. It truncates interface names, requires
labels "ifname:X" for each address, what limits the length again, fakes
interface names, which basically do not exists in the system [foo0:42;
there is only foo0 as valid interface] and [without label] supports only
one address per family on an initerface.
Finally, this causes bug reports and even feature requests about.
See e.g. https://bugzilla.novell.com/show_bug.cgi?id=639276
Other ifconfig implementations do not have these limits [AFAIR *BSD]
and adopted the output, which is AFAIK not standarized.
Every program, which is parsing the ifconfig output is assuming some
specific implementation [we have to maintain patches for it] and is
IMO broken per definition (-> use getifaddrs(3) or ip addr & co).
So from my point of view we can remove it from default install; every
bug report about it, is a step forward to fix/cleanup attic things...
Basically, we close bug reports about ifconfig truncation as INVALID
since ages, because it is considered obsolete.
Gruesse / Regards,
Marius Tomaschewski