Hello, I have a fast mirror, also available on IPv6.
Mandrake has broken mirror-checking scripts, their list of mirrors is > 1 day old, even though they claim they update it every 10 minutes. So don't bother to contact any of the four mirrors which are currently listed at their site.
And, of course, consider joining Mandrake club after downloading the distro.
I use this setup for ~1 year now - two VGA cards (one AGP, one PCI), two keyboards (PS2 and USB), two mice (PS2 and USB). You need to patch the XFree86 server. More info here.
The problem with external ISA bus (as opposed to ISA devices built in the southbridge) is that the southbridge knows which I/O ports are mapped on it, and can respond to a PCI transactions to this
port immediately. OTOH, when there is an external ISA bus, the southbridge has to propagate any I/O transaction to the ports unhandled by PCI devices to the ISA bus, which causes a delay in the whole
system. So it definitely does matter when you have
a system without ISA slot.
There are features which are missing on every
programming language I know (and which have the
ambitions to be used in high-performance computing):
Branch prediction: Modern CPUs (UltraSparc II, etc.) have two types of branch instruction. First one is attributed that in most cases the code continues after the instruction, and the other one is attributed that in most cases the code jumps
to a different address. The GNU C has __attribute__(branch) or something like
that. I think there should be two variants
of the if statement.
Memory barrier: The ability to finish the
previous store instructions, and serialize WRT write reordering and other CPUs.
And one more thing: IMHO the ability to use the
preprocessor is worth the additional complexity.
And it even allows to extend the language somewhat (conditional compiles, etc).
Now I wonder why this story has been posted on/.
a week after the release, while my own submission
on the same topic from last Monday (when the 7.2 beta has been released) has been rejected.
A friend of mine wrote an article entitled "Why do I prefer GPL over other free licenses"
I think it is worth reading, and he raises few valid points,
such as that GPL is extremely fair license (unlike BSD).
It was a bug in ProFTPd's usage of sendfile(). See the
discussion on linux-kernel yesterday. It should
be fixed on ftp.fi.muni.cz/ftp.linux.cz now. -Yenya
--
... have available the openssh-2.3.0p1 RPMs since
November 21 (actually a few hours/days later,
because Nov 21 is the build date of the package).
This is a long time.
Yes, you are right. I tried it a few years ago
on my P133 system with 32M RAM and the kernel
compile time was minimal for make -j4. With
both -j3 and -j5 the compile was slower. -Yenya
--
I was eagerly waiting for Linux 2.4 from the day I heard
somewhere it would support ipfilter.
Oh, this is the same confusion like in
Swansea NET-2 versus BSD NET/2. Linux never attempted to support ipfilter.
The core framework in kernel is called netfilter. IPtables are built on top
of netfilter (as is Linux virtual server, etc).
I don't know much about ipfilter, but I think
it is (at least partly) user-space solution.
Because we already had a fast kernel-space solution, I see no point in
moving back to the user-space.
With ipchains (2.2 kernel), I run router with
four 100Mbps ethernets (on an old Celeron 266)
and over 250 rules, and it works on full bandwidth. This is impossible with partial user-space solution.
You should probably use stat(2) instead of open(2),
as it is faster.
User-space queueing, or even long rule lists,
are slow. When you have to do
this on the high-performance routers (several
100Mbit interfaces), you will loose.
The right solution is use the ip rule
command, because it does not have to queue every
packet (it uses kernel routing cache). You can
probably even move it to the user space using
rt-netlink (in the same way as routing daemon
works) - just set up a blackhole route when the
stat(2) succeeds. This of course works as long
as you need to filter IP addresses, not ports.
As the FAQ says, and nevermind all the license blather, it's free
for personal use on ANY distribution.
Nope. It is not free, but they say they will
not waste time going after home users of
this product, but rather focus on the
(big) commercial subjects instead.
(1) that this about OpenOffice, not just StarOffice, and
(2) that the reason they can do this is because they are using the
Sun Industry Standards Source License (SISSL) instead of the GPL.
(OpenOffice is licensed under both GPL and SISSL, take your
pick.)
You are wrong. This is about StarOffice (5.2),
not OpenOffice (altough they use OpenOffice
sources to reverse-engineer the localization
info for StarOffice 5.2).
Isn't that the same reason
why recently Red Hat shipped with a development version of
GCC?
Yes and no. Everyone can take the same development
version of gcc (most probably even the same
RPM/SRPM from Red Hat Linux) and use it on another
distribution or incorporate it to another distribution.
What do people think they hope to accomplish with this silly
liscense? To prove that no-one will pay it any attention or what?
I should have probably added a bit more
to the background info. Here in the Czech republic, the most needed product for Linux
in the desktop area is the office suite,
and the one with the good (preferrably free
at least as in the free beer)
translation and localization support (printing in ISO 8859-2,
reading MS-Word documents, when MS does not
even use ISO 8859-2, but their own proprietary
encoding, CP 1250, etc). And StarOffice with
the SuSE add-ons is very near to these requirements. There are lots of newbie
users who are willing to listen, if you tell
them "Buy SuSE, because it gives you a Czech
office suite for free".
I view this as an unfair competition, but I doubt
this license is valid according to the Czech law
(IANAL, though).
Also remember that it will be available for all after May 31,
maybe Sun aren't allowing them to do it before.
But why do they release it for Debian and Slackware three months sooner, then?
I'm guessing we'll either have a clarification ("Sun's fault...") or a
retraction ("oops, some new guy in legal/marketing screwed
up...") within a couple of days.
No, the official statement (the link in the
story) is from Richard Jelinek, the general
manager of SuSE Czech (and no, he is not
a "new guy", he is with SuSE Czech from its
beginning).
Slashdot Mirror
They both support IPv6 (at least under Red Hat Linux).
-Yenya
This is probably the first release of RedHat Linux, which generates on my mirror less traffic, than a corresponding release of Mandrake Linux.
Disclaimer: I am the system administrator of the mentioned mirror site.
And, of course, consider joining Mandrake club after downloading the distro.
I use this setup for ~1 year now - two VGA cards (one AGP, one PCI), two keyboards (PS2 and USB), two mice (PS2 and USB). You need to patch the XFree86 server. More info here.
My mirror still has some 30 Mbps of free bandwidth, so if you are in Europe, you can try to download from it.
a 100Mbps mirror in Czech Republic, Europe can be found at ftp.linux.cz.
The problem with external ISA bus (as opposed to ISA devices built in the southbridge) is that the southbridge knows which I/O ports are mapped on it, and can respond to a PCI transactions to this port immediately. OTOH, when there is an external ISA bus, the southbridge has to propagate any I/O transaction to the ports unhandled by PCI devices to the ISA bus, which causes a delay in the whole system. So it definitely does matter when you have a system without ISA slot.
- Branch prediction: Modern CPUs (UltraSparc II, etc.) have two types of branch instruction. First one is attributed that in most cases the code continues after the instruction, and the other one is attributed that in most cases the code jumps
to a different address. The GNU C has __attribute__(branch) or something like
that. I think there should be two variants
of the if statement.
- Memory barrier: The ability to finish the
previous store instructions, and serialize WRT write reordering and other CPUs.
And one more thing: IMHO the ability to use the preprocessor is worth the additional complexity. And it even allows to extend the language somewhat (conditional compiles, etc).A week old "news" is not news anymore.
-Yenya
--
-Yenya
--
It was a bug in ProFTPd's usage of sendfile(). See the discussion on linux-kernel yesterday. It should be fixed on ftp.fi.muni.cz/ftp.linux.cz now.
-Yenya
--
-Yenya
--
-Yenya
--
Yes, you are right. I tried it a few years ago on my P133 system with 32M RAM and the kernel compile time was minimal for make -j4. With both -j3 and -j5 the compile was slower.
-Yenya
--
It was. Read it carefully (not the shortened version on the slashdot title page, but the full version under "Read more".
-Yenya
--
Oh, this is the same confusion like in Swansea NET-2 versus BSD NET/2. Linux never attempted to support ipfilter. The core framework in kernel is called netfilter. IPtables are built on top of netfilter (as is Linux virtual server, etc).
I don't know much about ipfilter, but I think it is (at least partly) user-space solution. Because we already had a fast kernel-space solution, I see no point in moving back to the user-space.
With ipchains (2.2 kernel), I run router with four 100Mbps ethernets (on an old Celeron 266) and over 250 rules, and it works on full bandwidth. This is impossible with partial user-space solution.
-Yenya
--
User-space queueing, or even long rule lists, are slow. When you have to do this on the high-performance routers (several 100Mbit interfaces), you will loose.
The right solution is use the ip rule command, because it does not have to queue every packet (it uses kernel routing cache). You can probably even move it to the user space using rt-netlink (in the same way as routing daemon works) - just set up a blackhole route when the stat(2) succeeds. This of course works as long as you need to filter IP addresses, not ports.
-Yenya
--
I hope I have clearly stated that it will be free in june and it will be free for Debian and Slackware in March.
OTOH, it is not free for personal use. They say they will probably not sue home users of this product. But it is not free.
-Yenya
--
Nope. It is not free, but they say they will not waste time going after home users of this product, but rather focus on the (big) commercial subjects instead.
-Yenya
--
You are wrong. This is about StarOffice (5.2), not OpenOffice (altough they use OpenOffice sources to reverse-engineer the localization info for StarOffice 5.2).
-Yenya
--
Yes and no. Everyone can take the same development version of gcc (most probably even the same RPM/SRPM from Red Hat Linux) and use it on another distribution or incorporate it to another distribution.
-Yenya
--
I should have probably added a bit more to the background info. Here in the Czech republic, the most needed product for Linux in the desktop area is the office suite, and the one with the good (preferrably free at least as in the free beer) translation and localization support (printing in ISO 8859-2, reading MS-Word documents, when MS does not even use ISO 8859-2, but their own proprietary encoding, CP 1250, etc). And StarOffice with the SuSE add-ons is very near to these requirements. There are lots of newbie users who are willing to listen, if you tell them "Buy SuSE, because it gives you a Czech office suite for free".
I view this as an unfair competition, but I doubt this license is valid according to the Czech law (IANAL, though).
-Yenya
--
But why do they release it for Debian and Slackware three months sooner, then?
I'm guessing we'll either have a clarification ("Sun's fault...") or a retraction ("oops, some new guy in legal/marketing screwed up...") within a couple of days.
No, the official statement (the link in the story) is from Richard Jelinek, the general manager of SuSE Czech (and no, he is not a "new guy", he is with SuSE Czech from its beginning).
-Yenya
--