[12:26] <Nafallo> BenC: hi! do you know of hand if grow for raid level 5 are implemented in the kernel yet?
[12:31] <BenC> Nafallo: grow by adding a disk?
[12:31] <BenC> if so, yes it is
[12:31] <Nafallo> nice. remember what kernel version? :-)
[12:31] <Nafallo> I have a user on #ubuntu-se that asks about it. running dapper :-P
[12:31] <BenC> I'm pretty sure edgy supported it
[12:31] <BenC> not sure about edgy
[12:32] <Nafallo> nice. I tell him to dist-upgrade and try then :-)
[12:32] <Nafallo> he wanted to compile 2.6.20 on dapper...
[12:32] <Nafallo> I guess thats a dumb idea? ;-)
[12:32] <kylem> no. it's fne.
[12:33] <kylem> booting it might be hard though.
[12:33] <kylem> 8)
[12:33] <Nafallo> lol
[01:47] <Nafallo> # CONFIG_MD_RAID5_RESHAPE is not set
[01:47] <Nafallo> :-P
[01:53] <BenC> Nafallo: I said the kernel supported it...I didn't say it was enabled :P
[01:53] <Nafallo> :-)
[01:53] <BenC> I will enable it for feisty though
[01:54] <Nafallo> yay!
[01:54] <Nafallo> so the next kernel will have it? :-)
[01:54] <BenC> yeah
[03:30] <zul> hey foks
[03:30] <Nafallo> hi zul 
[04:38] <kylem> BenC, did you mean to send that to ubuntu-dveel?
[10:19] <ivoks> anyone has an idea why linux-image packages in dapper and edgy depend on non-exsisting kernels? :)
[10:20] <cjwatson> ivoks: soyuz issue, being addressed
[10:20] <ivoks> thanks
[10:20] <cjwatson> ivoks: see #ubuntu-devel topic
[10:22] <ivoks> thanks once again :)
[01:47] <BenC> kylem: No, the guy that sent it to me Cc's -devel, and I just reply-to-all and sent the reply there to
[01:47] <BenC> he got moderated, I didn't :/
[01:48] <poningru> so in edgy there is a dependency error that came up today
[01:48] <BenC> I will go away soon
[01:48] <poningru> k
[01:49] <poningru> did you want the error?
[01:49] <poningru> just making sure
[01:49] <poningru> thanks
[01:50] <zul> what happened I just got up :)
[01:51] <BenC> poningru: You mean the linux-meta packages not being satisfied by a proper kernel to install...it's known
[01:51] <BenC> s/I/it/ too :)
[01:51] <BenC> I wont go anywhere
[01:51] <poningru> BenC: thats the one
[01:52] <poningru> thanks
[01:52] <poningru> :)
[01:52] <zul> heh by the number of bug reports there is alot people who use linux-meta :)
[01:52] <poningru> well its -generic-header and -generic for me
[01:52] <poningru> but yeah
[01:52] <poningru> the -meta packages
[01:53] <BenC> zul: linux-generic is installed by default on people's systems
[02:15] <MadMan2k> might I ask a non directly development related question?
[02:16] <MadMan2k> why do I get 500MB of modules when I compile the kernel myself using make-kpkg, while the official linux_image is only 70MB?
[02:19] <MadMan2k> I used oldconfig
[02:20] <Lure> MadMan2k: because you compile all varians (generic, i386, bigiron, lowlatency...)
[02:22] <MadMan2k> makes sense :)
[02:23] <MadMan2k> Lure: if I append "--arch generic" will it give me generic only?
[02:23] <Lure> MadMan2k: generic+lowlatency (afair), but I build directly on git
[02:32] <MadMan2k> is there a list of parameters with which the official kernels are being built?
[02:32] <MadMan2k> I basically only want to apply 2 patches and leave everything else as is.
[03:25] <kylem> BenC, ah ok
[03:30] <dholbach> hello
[03:31] <zul> hi dholbach 
[03:31] <pkl_> dholbach: hu
[03:31] <dholbach> from -6 to -7 it seems that my skge eth0 is not automatically enabled anymore - does anybody know what I could do to debug this?
[03:32] <pkl_> you means it's recognised (ifconfig eth0 reports it), but it is not configured?
[03:33] <dholbach> no, with -7, I have to   sudo ifconfig eth0 up   it myself)
[03:33] <dholbach> then I get a message in syslog that it got enabled
[03:35] <pkl_> so, once you've brought it up, it is okay, and it is configured?  How do you get your IP address?
[03:35] <dholbach> I run  sudo dhclient  
[03:40] <dholbach> if you run       wget http://daniel.holba.ch/temp/dmesg-{6,7}; diff -u dmesg-{6,7} | less       you will see the difference in what's happening
[03:40] <pkl_> ok, I was going to suggest a dmesg when you all disappeared...
[03:40] <dholbach> there's also the output of   lspci -vvnn   if you want that ( wget http://daniel.holba.ch/temp/vvnn-{6,7} )
[03:41] <dholbach> (just look for skge at the bottom)
[04:01] <dholbach> pkl_: you think I should file this as a bug?
[04:08] <pkl_> dholbach: it's definately a bug, but not sure in what though.
[04:09] <pkl_> dholbach: I'll have a look into how network interfaces are detected and brought up in Ubuntu.  What version of Ubuntu are you using?
[04:14] <dholbach> pkl_: feisty, 2.6.20-7-generic on amd64
[04:14] <dholbach> it's not terribly urgent, I can still use -6 - just didn't know if you need a bug report for that or how you proceed from there
[04:22] <pkl_> dholbach: I'm not sure either (having not been here long).  Personally I would raise a bug report.
[04:23] <dholbach> ok, I'll do that
[04:23] <dholbach> and I'll try on my i386 later on, if it happens there too
[04:30] <dholbach> bug 84218 - if you need more information, yell :)
[04:47] <BenC> dholbach: So the module loads, and the device is present in ifconfig -a, it just doesn't get configured?
[04:47] <BenC> dholbach: And going back to -6 (same userspace) works as expected?
[04:48] <dholbach> -6 works as expected - I didn't try   ifconfig -a    , plain   ifconfig   didn't show it
[04:48] <dholbach> I can reboot and try, if you like
[04:49] <BenC> yeah, without -a, you'll only see the ones that are up
[04:49] <BenC> if -a shows the device, then it's obviously working from the kernel side
[04:49] <dholbach> just un moment
[04:52] <pkl_> it comes up in the dmesg as recognised, but, it doesn't get configured.
[04:53] <kylem> try the sky2 module instead.
[04:53] <dholbach> it gets shown on   ifconfig -a  
[04:54] <dholbach> kylem: I modprobed it, but still it doesn't get configured
[04:55] <dholbach> BenC, kylem, pkl_: any more info I can get you?
[04:56] <BenC> dholbach: If it gets loaded and recognized, the kernel pretty much did it's job
[04:56] <dholbach> hum
[04:57] <BenC> no idea why booting one or the other would change that, unless some uevent is missing
[04:57] <BenC> do you use ifupdown, or network-manager?
[04:57] <dholbach> the latter
[05:01] <BenC> dholbach: Try ifupdown...if that works on both, then I suspect network-manager is being picky
[05:01] <dholbach> looks like the output of lshal changed completely
[05:01] <BenC> maybe the backend driver for skge needs to be updates for some trivial change
[05:02] <dholbach> maybe network-manager doesn't like that, nm doesn't seem to know about eth0 at all
[05:02] <kylem> oh.
[05:04] <dholbach> do you think it makes sense to attach the lshal logs and reassign to network-manager?
[05:04] <kylem> yeah.
[05:04] <dholbach> okay
[05:08] <dholbach> kylem, BenC, pkl_: thanks again - I reassigned it.
[05:08] <kylem> ok. cool.
[05:09] <kylem> dholbach, so it worked if you manually dhclient it?
[05:09] <dholbach> yes
[05:09] <kylem> ok. weird. ok.
[05:09] <dholbach> sudo ifconfig eth0 up && sudo dhclient      makes it happy
[05:09] <dholbach> (then I have to restart all the nm bits, so gnome knows it can download mails and browse porn, etc.)
[05:09] <dholbach> b0ng
[05:10] <kylem> hehe
[05:22] <MadMan2k> my self-compiled kernels (make-kpkg/ oldconfig) are always bloated at ~500MB. how can I make it compile as small as the official kernels?
[05:23] <pmjdebruijn> MadMan2k, erh, that's extremely weird
[05:23] <pmjdebruijn> MadMan2k, make-kpkg should produce just as small kernels
[05:24] <BenC> MadMan2k: if you are using the ubuntu config, disable CONFIG_DEBUG_INFO
[05:24] <BenC> we have it enabled so that we can get the linux-debug images
[05:24] <BenC> for oprofile and such
[05:24] <MadMan2k> that might be it
[05:25] <MadMan2k> gonna try it later - thx
[05:25] <maks_> hey guys have you ever had reports of freezing lenovo 300 laptop on ntpdate cmd
[05:25] <BenC> but we strip the vmlinux and *.ko's after the build
[05:25] <BenC> maks_: Not that I can recall, but check launchpad
[05:27] <BenC> kylem: do you have an ia64?
[05:27] <kylem> not accessible, no.
[05:28] <BenC> the new acx driver fails to build on ia64 because of the WLAN_* compat for it went missing...my i2k is dead for some reason
[05:28] <BenC> maybe I can dig it up from git
[05:29] <kylem> hmm.
[05:30] <BenC> got it
[05:35] <maks_> Benc: ok thanks
[05:39] <maks_> Bugs in Linux: No results for search ntpdate
[05:52] <BenC> kylem: I need you to set your umask on rookery to 664 at least for push's
[05:53] <kylem> eh? what happened?
[05:53] <BenC> kylem: And then run "chmod g+w . -R" in /srv/kernel-team/private/edgy-ubuntu.git
[05:54] <BenC> kylem: perms blocking me on that git tree
[05:54] <kylem> on .git/HEAD?
[05:54] <BenC> No, objects
[05:55] <kylem> uh..
[05:55] <kylem> well, i got a pile of -EPERM, but done.
[05:55] <BenC> weird
[05:55] <BenC> -r--rw-r-- 1 kyle kernel_team 2055 Jan 22 13:05 0244a8d6ba0bdda6f4c15484843d349a6ed68b
[05:55] <kylem> i don't see how you could be pushing the same objects as me.
[05:56] <BenC> did you have a 464 umask? :)
[05:56] <kylem> no.
[05:56] <kylem> 002.
[05:56] <kylem> er, 022
[05:56] <kylem> now it's 002
[05:56] <BenC> doh, writing inverted masks
[05:56] <BenC> I meant 002, right
[05:57] <kylem> eh?
[05:57] <cjwatson> ITYM 202
[05:57] <kylem> i'm confused.
[05:57] <cjwatson> Ben meant to ask if your umask was 202, I believe
[05:57] <BenC> when I said your umask, I wrote 664, I mean 002
[05:57] <kylem> i figured.
[05:57] <BenC> and then the 646 one
[05:57] <BenC> unable to write sha1 filename ./objects/95/30784fabf34b7c01b564f2ad3d2b14100097b6: Permission denied
[05:57] <BenC> that's what I got when pushing
[05:58] <kylem> must be a dir with the wrong perms.
[05:58] <BenC> not sure why it wouldn't let me create that object...maybe it'll work now
[05:58] <kylem> drwxrwsr-x 2 kyle kernel_team 4096 Feb  9 16:57 ./objects/95/
[05:58] <kylem> that's... right though.
[05:58] <BenC> yeah, that works now
[05:59] <kylem> sorry, i'm trying to clean my inbox a bit atm.
[05:59] <kylem> all the canonical private lists missed my procmail rules since i started, so i'm trying to sort a bit.
[05:59] <BenC> good luck...I hate cleaning my inbox more than I hate taking our the garbage
[05:59] <kylem> hehe.
[06:00] <cjwatson> ---Mutt: /var/mail/cjwatson [Msgs:9906 New:638 Old:1168 Post:9 Inc:52 79M] ---(threads/date)---
[06:00] <kylem> i finally worked up the balls to expire everything more than a year old from my synched maildir.
[06:00] <BenC> cjwatson: wow, my 19 emails seems a little humbled by that :)
[06:02] <kylem> hmm, i really don't need to be on activity@ anymore...
[06:03] <BenC> yay, two successful builds for 7.13
[06:03] <zul> gah..
[06:04] <zul> http://www.devxnews.com/article.php/3658001
[06:04] <zul> xensource is not looking into getting included in upstream
[06:05] <BenC> "legacy virtualization"?!?!
[06:05] <zul> heh..
[06:05] <BenC> "It lacks the benefits of para-virtualization performance enhancements that have been pioneered by Xen and are now being copied by VMware and Microsoft"!?!?!!
[06:05] <BenC> what kind of crack is this guy on?
[06:05] <zul> dunno
[06:05] <BenC> "Pratt also explained that Xen is no longer actively seeking inclusion in the mainline Linux kernel either."
[06:05] <BenC> lovely
[06:08] <kylem> i think that's kind o fbullshit.
[06:08] <BenC> it's a load of something, and it smells like my yard...bullshit sounds like a good guess
[06:08] <kylem> hah.
[06:09] <BenC> I think xensource likes Xen being hard and not integrated
[06:09] <zul> xen likes to make me bald
[06:09] <kylem> i think xensource wishes it was still a research project.
[06:09] <BenC> I think it still is
[06:10] <kylem> they seem to be trying very hard to make it a commercial product...
[06:21] <pkl_> cynically, a lot of the para-vitualisation work that has gone into the kernel for kvm/Lguest may conflict with Xen.  They've had a lot of resistance to how they do things on lkml, and may not wish to change their code to be more inline with kvm/Lguest changes.
[06:21] <pkl_> only a guess, but nothing else makes sense IMHO
[06:22] <BenC> paravirt isn't supposed to get rid of conflicts
[06:22] <BenC> err, is supposed to
[06:23] <BenC> Xen/KVM/VMI should all be able to live together if paravirt is implemented correctly
[06:23] <pkl_> yeah, which is why it doesn't make sense why they don't want to cooperate.
[06:24] <BenC> It's odd how they can use paravirt as a selling point for them and against the other implementations in the same sentence
[06:24] <zul> they are trying to cooperate
[06:24] <zul> http://lists.osdl.org/pipermail/virtualization/2007-February/002004.html
[06:24] <pkl_> but maybe they don't.  It may simply be sour grapes, they've been trying to get it accepted for two years, and a jhonny-come-lately has gone in with much less trouble.
[06:25] <kylem> the johnny-come-lately was... several MEGABYTES less delta to the core kenrle.
[06:26] <BenC> they say that kvm lacks paravirtualization, when that's simply not true...it's not in 2.6.20, but it's there
[06:26] <pkl_> heh, but they're probably coming from the direction that their several megabytes more delta was valid...
[06:26] <kylem> bullshit.
[06:26] <BenC> not to mention that kvm needs a lot less paravirt enhancement for the same performance, when the implementation is complete
[06:27] <kylem> they diverted huge swaths of core kernel code with #ifdefs instead of creating new abstractions.
[06:27] <kylem> they shouldn't have to look too far to see why they never did, and probably never will get merged as it stands now.
[06:27] <mjg59> Is the Xen code in Linux not just for the case where you want to run a paravirtualised Linux under Xen?
[06:27] <BenC> the xen para-virt domU patch was horribly large and touched way to much core kernel
[06:28] <mjg59> I almost got hired to do ACPI support in Xen
[06:28] <kylem> mjg59, i don't recall how much the dom0 stuff mucked with.
[06:28] <BenC> mjg59: No, up until recently, the Xen kernel stuff was dom0(hypervisor) and domU support all together
[06:28] <mjg59> Oh. Well, unsurprising that they lose.
[06:28] <pkl_> I'm not defending them, but I know how Cambridge university types think (having worked there).
[06:28] <mjg59> pkl_: I'm still a Cambridge university type :)
[06:28] <BenC> someone took the time to pull out the domU stuff with paravirt, but it was still a 500k patch, and touched > 150 files of core kernel
[06:29] <mjg59> Eurgh.
[06:29] <kylem> i think i'm ruined by being a kernel hacker first, and an academic second. i've been doing this since long before i started uni. ;P
[06:29] <pkl_> mjg59: really, you work at the Computer Laboratory?
[06:29] <mjg59> pkl_: No, I'm a geneticist
[06:30] <BenC> no one knows why mjg59 works on the kernel considering his field of study :)
[06:30] <pkl_> A lot of people at Cambridge have a superiority complex.  They may simply hate having come 2nd.
[06:30] <kylem> BenC, masochism was my best guess.
[06:31] <Nafallo> mjg59: do it! acpi is what I lack in XEN ;-)
[06:31] <mjg59> Nafallo: Nnngh.
[06:31] <kylem> pkl_, 2nd to what?
[06:31] <mjg59> Rusty.
[06:31] <pkl_> hm, second to kvm.
[06:31] <mjg59> Heh
[06:31] <kylem> oh.
[06:31] <mjg59> Ian Pratt is actually really nice
[06:32] <mjg59> Mind you, I say that about most people who've bought me booze
[06:32] <BenC> lol
[06:32] <pkl_> Yes, I used to know Ian Pratt, we went out drinking together in Cambridge several times each week for months whilst I worked there.
[06:33] <kylem> mjg59, that's a pretty good system
[06:33] <pkl_> Ian Pratt was finishing off his PhD, while I was working there as a post-doc.
[06:38] <pkl_> Ian Pratt went on to get a college fellowship, while I realised without a Cambridge PhD there was liitle chance of progression (and I too old to apply for a college fellowship).  I got my PhD from Lancaster University, which at that time was doing multimedia computing research comparable to Cambridge.  I decided to go to Acorn to design a new multimedia operating system for them.  They went bust 2 years later, the operating system got cancelled
[06:38] <pkl_> , and the rest is history.
[06:39] <mjg59> I want to be the bestest kernel hacker with a genetics PhD in the world ever
[06:39] <kylem> i think you already are, minus the phd part.
[06:39] <kylem> but if you'd stop wasting your time with acpi maybe you could get that done. ;P
[06:39] <mjg59> Bah.
[06:40] <mjg59> I've written 150 lines of perl today.
[06:40] <mjg59> That's most of a paper.
[06:40] <kylem> hehe.
[06:40] <kylem> why is perl so popular in biology?
[06:40] <pkl_> because fortran is old-fashioned?
[06:40] <mjg59> Because most genetics is just writing the right regexp
[06:40] <kylem> pfft. fortran is a fine language.
[06:41] <pkl_> I used fortran for a couple of things in 1985.  It was extremely popular at time in Physics and Engineering :-(
[06:42] <kylem> ancient code from goddard... in F77...
[06:42] <kylem> (this was like, in 2001, F95 had been around for... a while...)
[06:42] <pkl_> mjg59: BTW which department is in the New Museums site now the computer lab has moved?
[06:43] <mjg59> pkl_: Various people grabbed it
[06:43] <mjg59> Bits of archaeology, zoology, the computing service, materials...
[06:43] <pkl_> kylem: at least you've never had to learn/use COBOL.  COBOL 66 lovely :-)
[06:44] <kylem> heh. i've never even seen cobol code.
[06:44] <kylem> BenC, this looks bad:
[06:44] <kylem> Rejected:
[06:44] <kylem> UploadError escaped upload.process: File linux-source-2.6.17_2.6.17.1-11.35.dsc
[06:44] <kylem> as mentioned in the changes file was not found.
[06:44] <kylem> -----BEGIN PGP SIGNED MESSAGE-----
[06:45] <BenC> kylem: talk to pitti...might have something to do with the security>proposed pocket fixes
[06:45] <kylem> sigh.
[06:45] <BenC> maybe he didn't get all the files into the queue
[06:45] <zul> kylem: you dont want to...believe me...they still teach it at algonquin when i was there
[06:45] <BenC> Algonquin...I love saying that name
[06:45] <pkl_> cobol or fortran?
[06:45] <mjg59> My MBP should be turning up on Monday
[06:45] <zul> pkl_: both
[06:45] <mjg59> It seems to take years for Intel to get round to posting stuff
[06:45] <kylem> mjg59, it's hot. enjoy OS X.
[06:45] <kylem> :P
[06:46] <pkl_> yuck, at least fortran is concise.  Cobol is a managers idea of a programming language, things are grouped together in parapgraphs.  And each line of code is a sentence.  No joke...
[06:47] <kylem> yow.
[06:47] <zul> pkl: heh i still remember the four dvisions of cobol thats about it
[06:47] <pkl_> Perform para1 thru para8 varying counter by 1 until counter equals 10.  or something like that is a for statement.
[06:51] <pkl_> yeah, the "initialization division".  Try typing that from the UK, and more often than not you type "initialisation division".  On a batch processing system, when compiles are added during the day, and run at night in batch mode, you can wait a whole day to find you've used a 's' rather than a 'z' :-(
[06:51] <mjg59> BenC: 84238 - we should get 2.6.17 removed
[06:54] <cjwatson> the computer lab was never all that much of the new museums site anyway
[06:55] <cjwatson> I mean, the tower and stuff, yeah, but I kept finding new departments on the NMS that I'd never previously heard of
[06:55] <cjwatson> the place is a TARDIS
[06:56] <cjwatson> huh, 2.6.17 isn't gone already?
[06:57] <mjg59> Well, either that or the user is awfully confused
[06:57] <cjwatson> BenC: ack linux-source-2.6.17 removal from feisty?
[06:57] <pkl_> yeah, I meant the bit of the NMS that the lab occupied.  All the 'junk' than had accumulated had to go somewhere, the cap computer was still stuck in a corridor when I was there.
[06:57] <BenC> cjwatson: Sure, was just looking into filing a bug for it :)
[06:57] <cjwatson> it does seem like a bug that we break 2.6.17's initramfs with feisty's tools, though
[06:57] <zul> cjwatson: can you remove xen-source-2.6.17 as well?
[06:57] <cjwatson> that will surely cause upgrade pain
[06:58] <cjwatson> zul: what's current in feisty?
[06:59] <zul> 2.6.19
[06:59] <cjwatson> oh
[06:59] <cjwatson> he ran mkinitrd
[07:00] <cjwatson> not mkinitramfs
[07:03] <cjwatson> zul: oh, I was confused by the source package being just 'xen-source' now
[07:03] <cjwatson> BenC,zul: done
[07:03] <zul> cjwatson: thanks..
[07:07] <BenC> cjwatson: thanks
[07:21] <fvaler> bochs?
[07:42] <kylem> mjg59, fwd'd the mail from alan to you and ben. i'll try cooking a patch up on monda.
[07:47] <pkl_> kylem: is that email about the para_acpi driver?
[07:47] <pkl_> para -> pata
[07:48] <kylem> no. HPA.
[07:48] <mjg59> kylem: It's a pretty tiny amount of code
[07:48] <pkl_> okay, the protected area stuff?
[07:49] <mjg59> kylem: I suspect that the difficult bit is just getting it into a state where the libata guys will take it
[07:49] <mjg59> pkl_: Yp
[07:49] <pkl_> can you forward it me as well then?  I am apparently part of the kernel-team now?
[07:50] <kylem> email?
[07:52] <pkl_> ah, that's what you said it was.
[07:52] <kylem> i was asking for your address, but i went and fetched it anyway.
[07:53] <pkl_> oh good :-) phillip.lougher@canonical.com would have worked.
[07:54] <cjwatson> BenC: could you have a look at bug 84094? symlinks seem to be buggered after installation
[07:57] <pkl_> kylem: Diolch yn fawr... (that's thanks very much BTW in Welsh, as its about Alan).
[07:59] <BenC> cjwatson: checking...
[08:04] <fvaler> the linux kernel is more often than not in C or c++?
[08:04] <fvaler> C isn't it
[08:04] <mjg59> All in C
[08:05] <pkl_> C and a small amount of assembler
[08:05] <fvaler> Excellent
[08:05] <fvaler> So only the windows kernel is in c++?
[08:05] <fvaler> atleast large chunks of it are
[08:05] <pkl_> No, Symbian (used on mobile phones) is C++
[08:06] <fvaler> which would you guys p refer to use
[08:06] <pkl_> Or a bastardised variant anyway.
[08:06] <fvaler> c/c++ as far as the kernel goes?
[08:07] <pkl_> C is lower-level, allows more exposure to the underlying architecture, is more efficient.  I wouldn't use C++ for anything personally.
[08:08] <kylem> eh? anything you can do for a kernel in C, you can do in C++, you can even use almost all C++ language features in a kernel...
[08:08] <fvaler> ok personally that sounds fine
[08:08] <fvaler> kylem: exactly
[08:08] <kylem> that said, i hate C++ with a firey burning hatred.
[08:08] <fvaler> why?
[08:09] <kylem> Mac OS X is written mostly in C++.
[08:09] <fvaler> I really dont understand this
[08:09] <fvaler> large chunks of windows is written in c++ as well
[08:09] <kylem> because for a kernel, it's overkill unless you are a really lazy person.
[08:09] <kylem> the only real benefit to Linux would be implicit "this" pointer for function pointers in a struct
[08:09] <fvaler> BUt it doesn;t lack anything C can do it can't
[08:09] <kylem> so you don't have crap like, sb->someop(sb, ...);
[08:10] <kylem> g++ is slower than dragging source lines through molasses anyways.
[08:11] <pkl_> this is a religous way, arguments for and against C and C++ can be made.  I've never liked C++ as it encourages abstraction and information hiding when in the kernel you don't anything that bloats or makes it slower.
[08:11] <pkl_> way -> war
[08:12] <kylem> c++ also encourages overuse of function pointers, which are pretty nasty on most processors.
[08:13] <fvaler> kylem: actually i dont think you can use c+= for kernel
[08:13] <fvaler> neither on windows or linux
[08:13] <fvaler> The difficulty is controlling linkage for things such as template instantiations
[08:13] <kylem> you can.
[08:13] <kylem> you have to disable rtti.
[08:13] <fvaler> http://www.microsoft.com/whdc/driver/kernel/KMcode.mspx#EDE
[08:14] <fvaler> C++ has RTTI.
[08:14] <mjg59> The only C++ I'm familiar with in the Linux kernel is a huge pile of junk that the linuxant drivers wedge in
[08:14] <kylem> ...
[08:14] <fvaler> C++ without RTTI isn't C++. Furthermore, that's far from the only problem.
[08:14] <pkl_> C++ is probably better for large projects with clear separation between components developed by multiple teams.  Often the gain in development is compensation for the (potential) loss of performance
[08:14] <kylem> ok, you're either dense, or an idiot.
[08:15] <kylem> one way or another, you're wrong, and this is off-topic.
[08:15] <fvaler> kylem: i'm sorry?
[08:15] <fvaler> C++ has RTTI
[08:15] <kylem> -f-no-rtti
[08:15] <kylem> welcome to my /ignore.
[08:15] <fvaler> Woah
[08:15] <pkl_> yes, this off-topic
[08:16] <fvaler> Offtopic maybe but we're talking about the kernel actually
[08:16] <fvaler> Just in C++
[08:16] <kylem> kickban someone?
[08:16] <fvaler> Peopple who disagree with you are idiots always?
[08:16] <kylem> http://netlab.ru.is/exception/LinuxCXX.shtml
[08:17] <mjg59> fvaler: Two issues:
[08:17] <mjg59> fvaler: You're wrong that C++ can't be used in the kernel. It's just that if you do it, nobody will ever speak to you again.
[08:17] <mjg59> fvaler: And secondly, this channel is for discussion of development of the Ubuntu kernel, not generic kernel issues
[08:18] <fvaler> I understand but name calling over something he disagrees with is preposterous
[08:18] <fvaler> Not mature
[08:18] <fvaler> let's read his link anyway
[08:18] <pkl_> fvaler: I've never treated anyone as an idiot for simply disagreeing.
[08:18] <mjg59> Well, it's not a disagreement. It's a statement of fact.
[08:18] <mjg59> But that's unimportant.
[08:18] <fvaler> Ok
[08:19] <fvaler> kylem: That research is about changing the linux kernel to support C++.
[08:20] <fvaler> Note the implication that it doesn't support it.
[08:20] <fvaler> It still doesn't.
[08:20] <kylem> it never will. because it's a shitty language. now GO AWAY.
[08:20] <kylem> take it to #kernelnewbies.
[08:20] <zul> ok enough...this is totally offtopic
[08:22] <fvaler> I'm at kernelnewbies
[08:22] <fvaler> And they agree with what I said it simply doesn't
[08:23] <zul> ENOUGH
[08:24] <BenC> fvaler: Basic idea is that whether or not C++ will work, or can work in the linux kernel is completely irrelevant to this channel, and most people here probably feel that the topic is more trolling than anything
[08:25] <fvaler> BenC: I gave up wrt to the topic and pretty much done
[08:25] <kylem> fvaler, look, i'm sorry, but i /HAVE WRITTEN/ linux kernel c++ code before. classes and all. that's clearly C++. i don't want to argue anymore.
[08:25] <fvaler> That paper doesn't say anything
[08:25] <fvaler> The paper I gave you is about using C++ in ways that are compatible with the kernel; the point is that C++ isn't supported.
[08:25] <fvaler> But yeah ok.
[08:26] <kylem> maybe not fully, but you can certainly get by.
[08:29] <fvaler> Not for the sake of arguing but lots of people have written C++ code for a kernel, and even more have compiled C code with a C++ compiler for a kernel
[08:29] <fvaler> Dude my only point is it has no bearing on whether C++ is a language supported by the kernel.