Day 2: Keynotes: Past, Present and Future of Linux
Tove, Linus' wife and their two girls where also there to enjoy the
talk
This is obviously the biggest keynote, and while I counted about
3500 people, there were a few hundred less this time, the room was still full
and I estimate the number of attendies to a little more than 3000.
As Linus just loves doing slides and preparing talks , he
gave a quick overview on the state of affairs, and then switched to Q&A
- 2.2.11 came out the night before his talk, but as he explained, it's
basically a handoff to Alan Cox, who maintains stable releases while Linus
works on the unstable tree. The plan is for 2.2.11 is to last until 2.4
(famous last words ).
- Feature freeze, and code freeze in a couple of months hopefully with a
release of 2.4 in december. As explained previously, like at the
BALUG talk, the 2.1 tree lasted way too long so
he doesn't want this to happen to the 2.3 branch
- 2.4 will have USB, I2O, careful resource allocation (for ISA PnP and
PCMCIA)
- Capability levels for processes
- 2.4 will not have a journaling FS, although it can be added by vendors in
2.4 later.
- As for Merced, Hopefully 2.4 will be released before the Merced NDA is
lifted, so no IA64 support in stock 2.4.
- Special kernel support for clustering will also have to wait for the next
release.
Then, Linus switched to the Q&A section. Some of the answers he gave were:
- Firewire support is not very well documented and used yet, so it's a bit
behind because of that.
- Maintainers are in charge of drivers, and he mainly cares about the core
so he doesn't look too closely to driver code submitted to him. As for
indentation and things like that, as long as he doesn't have to deal with
the code to closely (like for drivers), it's ok but otherwise he will
usually reject the patch.
- ACPI support is being worked on and it's more important than APM because
of its better capabilities.
- As for devfs, he doesn't like the naming of the devices all that much but
it's something he is looking at.
- Current scaling is up to 4-8 CPUs, but to go above like SGI is planning to
do,
it will probably be a different kernel.
- For DVDs, the main problem is the encryption and the secret algorithms
involved. The solutions are full hardware decryption and a commercial
software decoder. Both are currently being worked on.
- For realtime, the kernel is being improved to be a bit more deterministic,
but without
RT-Linux
where Linux is merely a low level RT task it won't be hard realtime. Linus
also says that many people who think they need hard realtime really don't.
- The ISDN code has been updated and the developers have now understood
(after a few messages on the linux kernel list) that they need to provide
patches more often and be better at staying in sync.
- For hotswap PCI, the problems are the same than what you face with ISA PnP
and PCMCIA, so it's being taken care of at the same time. While hotswap PCI
may not be in 2.4, the infrastructure will definitely be there.
- Rusty Russell asked about including kernel debugging patches. Linus, as
it's well known, doesn't really like the idea of kernel debugging in the
stock kernel as kernel debuggers may hide the real bug, and may lead
to people fixing the bug improperly.
- The alsa sound drivers will hopefully be in 2.4, and if not in the next
stable release.
- Linus is for kernel httpd support, as the current existing code is
really clean and logical. It should be looked at on a case by case basis.
Of course, inclusion in the kernel is not for much more complex things like
the samba server.
After that, Linus introduced Patrick McGovern, founder of IDG, who then had
Richard Stallman come on stage in order to give him the IDG/Linus Torvalds
Community Award for the Free Software Foundation.
RMS gave the usual GNU/Linux speech (i.e. because the basic unix utilities
(including the file and binutils), as well as a non trivial piece, the C
compiler, come from the GNU project, Richard is asking everyone to call Linux,
GNU/Linux in order to give credit to GNU).
My personal take on this is that while the GNU project definitely deserves
credit for offering a lot of the initial blocs that Linux is built on, someone
found that only about 10% of SuSE Linux was based on such software, so many
many other people deserve credit, the whole OSS community for that matter.
Therefore insisting every single time that you should use GNU/Linux gets real
old real fast... As far as I'm concerned RMS is more fun when he sings his
"Join us now and share the software"
song
If you feel you have to Email me to tell me how wrong you think I am, be my
guess, but I won't take the time to answer, this is a well known subject that
some people disagree on.
The good news in all this is that Linus' two girls didn't get bored, and
actually had quite a good time
Email
Link to Home Page
99/03/13 (03:42): Version 1.0