Guido van Rossum pretty much wanted a BSD-style license, but CNRI didn't agree. AFAIK, everybody in the BeOpen Pythonlabs team wants a GPL-compatible license.
The only objection of RMS is against clause 7, the "State of Virginia"-clause. The strange thing is, this clause may not be applicable outside the US, so the license may be GPL-compatible outside the US after all. (Only lawyers can sort this out).
(See http:// mailman.beopen.com/pipermail/license-py20/2000-Sep tember/000073.html.)
Perhaps some European lawyer could look at the clause (SuSE? RedHat Europe?). Maybe CNRI could be persuaded to drop the clause if it's void in the rest of the world (good) (or maybe they would append a ton of legalese to the license (bad)).
I was pretty shocked by next article that an OpenBSD-user showed me. (I'm a Linux user, thank you). You can read it here.
From: Theo de Raadt comp.unix.bsd.openbsd.misc
...
> I'm just > curious to know if anyone has broken into an open source system because it > was open source.
Linux is the best example of this, there are many examples. As the system being attacked, that is -- even if their source was not being analyzed earlier. Funny thing is, (especially around two years ago) it was a case of _us_ finding the holes, fixing them in OpenBSD, telling the BUGTRAQ mailing list, and then crackers writing exploits and using them on _other_ operating systems. (I guess that is distributed and applied;-)
Sometimes, as in the case of the recent RedHat lpd security report, years elapse. Let's look closer at what happened:
I wonder why something like BSD's ports collection doesn't exist in the Linux World. I frequently find myself hunting for a.src.rpm in order to try to fix something or at least have a glance at the code, but
not all mirrors mirror the SRPMS.
most SRPMS won't build when you're not root.
(I don't like tarballs because I tend to lose track: what's installed, where it is installed, what file belongs to which package... Plus I like having the patches that integrate the software in the distro.)
(I know there are other package managers but AFAIK they don't support source packages as well)
The main problem with IPv4 that IPv6 is trying to solve is a lack of address space. By using IP masquerading, that problem can be alieviated indefinately, at the cost of increasing the lag time. You get one IP address, which you then use IP masquerading to get up to 2^32 (minus oddballs like 127.*) addresses internally.
What if the masquerading machine fails?
Masquerading introduces a single point of failure, as there is no way to balance the load between different masquerading servers. Since every node uses the masquerading machine's IP-address, you can't tell it to switch to another machine without losing all established connections. The only solution is to use different subnets with different masquerading gateways, so only half of your network loses its net-connection when a masquerading machine goes down.
And if that's not enough addressing for you, you can run IP masquerading on each machine of your internal network, increasing the layers indefinately.
It doesn't make any sense to masquerade as a private IP. Just use a private class A network.
Maybe the 6bone is working ok, but a bunch of the proposed standards already seem kind of dated, and some parts of the proposal seem frankly dumb. Tying IPv6 addresses to link-layer addresses (and using half the 128 bits for it!) is dumb - it makes it very difficult to switch hardware around without having to wait for DNS changes to propagate, and there are plenty of other reasons to object to it too.
IMHO, it's not necessary to use the MAC-adress. Just as long as the machines can find each other through Neighbourhoud Discovery, everything's fine. In some cases it's not even possible to use the Interface Identifier because it doesn't exist: PPP-connections,...
RFC 2373 "IP Version 6 Addressing Architecture" cleary states:
An unicast address or a set of unicast addresses may be assigned to multiple physical interfaces if the implementation treats the multiple physical interfaces as one interface when presenting it to the internet layer.
It's quite annoying when you can't get on the net because the DHCP-server crashed.
Slashdot Mirror
Guido van Rossum pretty much wanted a BSD-style license, but CNRI didn't agree. AFAIK, everybody in the BeOpen Pythonlabs team wants a GPL-compatible license.
p tember/000073.html.)
The only objection of RMS is against clause 7, the "State of Virginia"-clause. The strange thing is, this clause may not be applicable outside the US, so the license may be GPL-compatible outside the US after all. (Only lawyers can sort this out).
(See http:// mailman.beopen.com/pipermail/license-py20/2000-Se
Perhaps some European lawyer could look at the clause (SuSE? RedHat Europe?). Maybe CNRI could be persuaded to drop the clause if it's void in the rest of the world (good) (or maybe they would append a ton of legalese to the license (bad)).
I was pretty shocked by next article that an OpenBSD-user showed me. (I'm a Linux user, thank you). You can read it here.
From: Theo de Raadt
comp.unix.bsd.openbsd.misc
...
> I'm just
> curious to know if anyone has broken into an open source system because it
> was open source.
Linux is the best example of this, there are many examples. As the ;-)
system being attacked, that is -- even if their source was not being
analyzed earlier. Funny thing is, (especially around two years ago)
it was a case of _us_ finding the holes, fixing them in OpenBSD,
telling the BUGTRAQ mailing list, and then crackers writing exploits
and using them on _other_ operating systems. (I guess that is
distributed and applied
Sometimes, as in the case of the recent RedHat lpd security report,
years elapse. Let's look closer at what happened:
http://www.pr ogressive-comp.com/Lists/?l=bugtraq&m=947550717304 74&w=2 5 60&w=2 5 70&w=2
http://www.pr ogressive-comp.com/Lists/?l=bugtraq&m=94731014106
http://www.pr ogressive-comp.com/Lists/?l=bugtraq&m=94755201131
And then read
http://www.pr ogressive-comp.com/Lists/?l=bugtraq&m=947699382089 89&w=2
And pay special attention to the original bug report. That's October
of '97.
http://www.nai.com/ nai_labs/asp_set/advisory/20_bsd_lpd_adv.asp
Perhaps a new distro would be a better idea
I wonder why something like BSD's ports collection doesn't exist in the Linux World. I frequently find myself hunting for a
(I don't like tarballs because I tend to lose track: what's installed, where it is installed, what file belongs to which package... Plus I like having the patches that integrate the software in the distro.)
(I know there are other package managers but AFAIK they don't support source packages as well)
What if the masquerading machine fails?
Masquerading introduces a single point of failure, as there is no way to balance the load between different masquerading servers. Since every node uses the masquerading machine's IP-address, you can't tell it to switch to another machine without losing all established connections. The only solution is to use different subnets with different masquerading gateways, so only half of your network loses its net-connection when a masquerading machine goes down.
And if that's not enough addressing for you, you can run IP masquerading on each machine of your internal network, increasing the layers indefinately.
It doesn't make any sense to masquerade as a private IP. Just use a private class A network.
IPv6 is way too scary to actually work :)
Scary? It's actually ment to be invisible :-)
IMHO, it's not necessary to use the MAC-adress. Just as long as the machines can find each other through Neighbourhoud Discovery, everything's fine. In some cases it's not even possible to use the Interface Identifier because it doesn't exist: PPP-connections,
RFC 2373 "IP Version 6 Addressing Architecture" cleary states:
It's quite annoying when you can't get on the net because the DHCP-server crashed.
Are the *BSD's compatible? Can I use some OpenBSD-package on FreeBSD?
Bitkeeper is not GPL'ed, so are they really going to use it?