Re: Top ten interesting broken packages (Volunteers?)
Hello On 08/28/12 07:30, Jelle Hermsen wrote: Great, does it also install well? I managed to get racket to bmake build fine, but when I bmake install it throws up errors during the compilation of the documentation. I did get racket-textual to build and install well on df32, and it runs fine. I haven't tried df64 yet. However, having Racket with a working GUI would be much better of course! I would love to see Larcency in pkgsrc too. On further investigation, my build was racket-5.2 from my last round of pkgsrc builds (~approx 1/2012 pkgsrc tree), and not the latest 5.3 Indeed the graphical 5.2 version did build completely, including documents, packaging, etc - I've put my (patched) pkgsrc lang/racket subtree in: http://leaf.dragonflybsd.org/~cat/racket-5.2/ in case any of the patches might be of assistance and am currently uploading the actual package to the same directory As for 5.3, I'll be working on another round of local-bulks in the coming weeks, so if you don't manage to get it sorted out by then, I'll review and take a crack at the 5.3 version whenever I get to that. If you're itching for a copy of my larceny build or notes let me know and I can post those for the time being. In other scheme porting misc, there is a patch in ypsilon bugs for DF support (#150), and chibi scheme now has DF support as well (bug #122) Also hope to get these all pkgsrcified eventually as well. Cheers, - Chris
Re: Top ten interesting broken packages (Volunteers?)
On 08/13/12 07:08, Jelle Hermsen wrote: I can take a look at fixing Racket. Should I test on both df64 and df32? Doh - sorry for sleeping on the list - have been switching jobs and my DF time has been languishing - If this hasn't been done I have a patch for racket32 lying around[1] - just haven't gotten around to coordinating / sharing / etc. will up-the-ante on this and try to get it available this week. Cheers, - Chris .. [1] side note for any schemers: larceny bootstrapped successfully from racket as well - will try to get this automated a-la clisp-sbcl bootstrap setup as a pkgsrc-wip package one fine day
Re: Fwd: df64 pkgsrc 2012Q2 DragonFly 3.0/x86_64 2012-07-24 11:35
On 08/07/12 16:16, Justin Sherrill wrote: It was somewhere around 2500, I think, and then the filesystem filled up. It was writing to disk faster than it was pruning, I think, ... no idea about the build setup - so maybe you're doing this already but for my local builds I tend to setup an hourly hammer cleanup (with appropriate hammer retention config) for my ${WRKOBJDIR} pfs - or just setup a UFS partition for same - there is simply too much churn on a big build to do otherwise and it's not like I really need the hammer historical features for the build state 2c, for whomever, fwiw Cheers, - Chris
Re: binary packages for 2012Q1
On 05/08/12 14:17, Pierre Abbat wrote: The problem is that py26-twisted is required for some packages, and it conflicts with py27-twisted, which is apparently required for some packages that it will upgrade later. You might review the python options from iirc lang/python/buildlink3.mk - there are some knobs for 'allowed / preferred, etc' python versions which may/may not work to help with those packages. Cheers, - Chris
pkg/46340: lang/openjdk7: DragonFlyBSD port
FYI I've finally gotten around to submitting a bug r.e. this port / emailing maintainers / tech-pkg@, etc. So.. hopefully the next mail will advise that it has been committed, and everyone can then begin to stare at NoClassDefFound stack traces as they try and configure all of their fave java goodness (hmm) to run on their fave OS. That being said, early adapters are still welcome to review previous users@ posts on how to get this running at present before it has been committed. and thanks again to FTigeot for blazing the trail on this Cheers, - Chris
Re: sound
On 03/21/12 03:28, Chris Turner wrote: Note: not everything builds / runs 100% with all features on DragonFly there, so keep this in mind expect to muck around Forgot to mention that this can be a good way to learn about various system specific sound issues and maybe fix a few packages along the way.. Cheers, - Chris
Re: scroll wheel doesn't work
On 03/14/12 06:04, Pierre Abbat wrote: Here's my latest try: I ran man psm and found that there are three operation levels which return different formats. In level 0, the mouse driver does not return scroll wheel information; in level 1, it does. This is probably overkill. Also, moused has an '-l' flag for this For reference - alot of the mouse documentation is somewhat 'cruft' from the MS/DOS PC-AT era and various nonstandard serial port drivers - for the most part things just 'work' for 'modern' (e.g. Pentium+/win95+ era) ps2/USB mice in my experience. However, some KVM's do 'interesting' things with port data, so you'll probably want to post a dmesg to see if anything obvious stands out, and try running from the console with -f and -d flags will probably be of use here as well - for example - a test session for me shows 'Z' axix activity in the 'dz' field here: $ grep 'dz.*1' /tmp/mouse-log moused: activity : buttons 0x dx 0 dy 0 dz -1 moused: activity : buttons 0x dx 0 dy 0 dz 1 I misadvised earlier in this thread r.e. the '-z' flag and apologize - at the time I was not using moused have switched back for testing here - my setup (moused, ps2 mouse, scroll OK) is as follows: $ dmesg |egrep '(psm|ums)' psm0: current command byte:0065 psm0.atkbdc0.acpi0.nexus0.root0 psm0: PS/2 Mouse [tentative] irq 12 on atkbdc0 psm0: model IntelliMouse Explorer, device ID 4-00, 5 buttons psm0: config:, flags:0008, packet size:4 psm0: syncmask:08, syncbits:00 psm0: PS/2 Mouse [attached!] irq 12 on atkbdc0 ^ note - much of the debug info is avail here in a verbose boot. $ pgrep -fl moused 934 /usr/sbin/moused -3 -p /dev/psm0 -t auto $ sed -ne 39,45p /etc/X11/xorg.conf Section InputDevice Identifier Mouse0 Driver mouse Option Protocol auto Option Device /dev/sysmouse Option ZAxisMapping 4 5 6 7 EndSection Also again, switching the Xorg 'device' to the raw device may help if you don't care about console usage. Good luck!
Re: gpg-agent was Re: system rebooted for no apparent reason
On 03/10/12 19:28, Pierre Abbat wrote: I installed pinentry-qt4 and added the lines to the conf files, but I'm not sure what to do about X. The instructions assume that I start KDE by logging into the console and typing startx. I use kdm, and I'd like it to work whether I log into KDE (my usual) or XFCE (something doesn't work, but it's probably fixed in the new quarterly). Where do I put the gpg-agent command? You'll have to either customize your KDM session scripts for KDE XFCE to launch this (see the KDM docs), start it from a terminal with an active gpg-agent, or hack up a KMail start script which starts it first and connect your menu items to that instead of KMail directly. Xdm has a hook to run ~/.xsession rather than changing the system wide scripts - this might work here - or if not there might be some KDM analogue. Not a KDE user so not 100% Basically, whatever process starts KMail will need to have the environment variable settings generated by the gpg-agent, wherever that falls in the sequence of inherited environment vars Cheers, - Chris
Re: system rebooted for no apparent reason
On 03/10/12 07:49, Pierre Abbat wrote: There was no kernel dump. How do I figure out what happened so that I can file a bug report? I'm not good at debugging these kinds of crashes myself - perhaps someone has some tips. However - if your signing setup in any way involves 3.0+ and a gpg-agent running as root - you might want to update from source to latest master-or-3.0 patch branch before digging since there are known issues with gpg-agent running as root that have been fixed. Also - if your KMail was running pinentry-curses - you might want to change that to one of the graphical password prompts to see if that changes things.. See also: http://freebsd.kde.org/howtos/gnupg-kmail.php which should fairly similar and discusses this. Cheers, - Chris
Re: Install DragonFlyBSD on 32 MB RAM
On 03/06/12 01:28, Edward M. wrote: On 03/05/2012 10:25 PM, v...@ukr.net wrote: What's the mistake and how can I fix it? try adding it to: /usr/pkg/etc/mk.conf Indeed - you probably also want to choose the more specific PKG_OPTIONS.bind98 version as the 'default options' setting will turn these things off in all packages.
Re: Install DragonFlyBSD on 32 MB RAM
On 03/06/12 05:24, v...@ukr.net wrote: On Mon, 05 Mar 2012 23:28:59 -0800 But maybe it is worth fixing this page: http://www.dragonflybsd.org/docs/newhandbook/pkgsrc/#index7h3 as well, because that was the place where I learned about the '/etc/mk.conf/ file? Fixed - thanks for the report. Pkgsrc used to default to /etc/mk.conf a while ago but now uses /usr/pkg/etc/mk.conf. The documentation wasn't yet updated. By the way, is it a good idea to disable the threads' option when building the applications on the machine with such a small amount of memory as mine has? I'm not sure about the exact meaning of this option - where can I get any description of it (as well as any other options)? Generally I wouldn't mess with the options unless you're 100% sure you want to disable it. To find out what it does, check the Makefile or options.mk file in the package directory and if it's not well documented, you'll need to see how the options affect the build process and dig in the package's source to see what this implies. The pkgsrc guide is quite good: http://www.netbsd.org/docs/pkgsrc/ and aside from this and the makefiles, perhaps check the CVS history. see also: http://pkgsrc.se for a nice web interface And also, the question about kernel build failure remains - I would appreciate any input on that case as well. Can you build GENERIC with a stock /etc/make.conf? If not, checkout a fresh source tree try a clean build. if so - you'll need to be making your kernel config file available and any build flags used if you want someone to try to fix. This particular error looks to be related to changing the VISUAL_USERCONFIG and can be found fairly simply by checking the file in question at the line where the build fails - the entire function is ifdef'ed out based on this option.. so - looks like this option is required most likely (I'm not 100%) Really - if you're making a custom kernel config and are changing options without checking what they do in the source tree - expect things to fail both in the build and while running, and expect to get your hands dirty - which means reading the source and finding out what the options do to the code. In other words - by continuing down the (unsupported) low memory path, with a custom kernel config, you've essentially become the 'low memory tester / developer' - and so will need to do 'tester/developer like things' if you want to make progress.. Hope this doesn't come across too harshly - the goal isn't to yell but to try and describe what I think is the reality on this thread for whatever that's worth. Cheers good luck! - Chris
Re: Install DragonFlyBSD on 48 MB RAM
On 03/02/12 10:51, v...@ukr.net wrote: After that my new system booted properly into multi-user mode and now I can connect via SSH to it and more or less do what I want. :) Glad it worked for you! P.S. After the installation I ran 'make pkgsrc-create' like DragonFlyBSD online documentation suggests, and it has been running for about 2 days already. :D With this low of memory, I would strongly suggest either simply using the public binary packages, or setting up another system or VM as a build machine to build your own binary packages. This will also come in handy if you want to do a system upgrade from source, since you probably won't be able to build the system on that low of memory without waiting quite a while due to swapping. Cheers Good Luck
Re: XScreensaver Tricks?
On 02/20/12 15:51, Chris Turner wrote: Any ideas? This was due to the crypt(3) changes committed in January which are not forward compatible with new passwords - so if you are running old binaries that are having problems authenticating against new password hashes, this is probably related the binaries should be rebuilt. Cheers
Re: Install DragonFlyBSD on 48 MB RAM
On 02/27/12 11:10, v...@ukr.net wrote: Now when I try to run 'installer', it shows me a blue screen with a dragonfly, but after a couple of seconds it vanishes and gives me the following message: Perhaps someone can chime in with an installer tip - However, given this is a non standard and sort of 'expert' install, you might review the 'INSTALL' file under / from the shell and try a manual install - its actually fairly straightforward - steps being roughly: - partition disks - install bootloader - create mount filesystems - copy over files - remove some installer specific files from the copies - reboot disclaimer: some of the disklabel / partitioning stuff is a bit out of date w/r/t disklabel64 / slice numbers, etc. I keep meaning to send a patch.. but anyhow.
XScreensaver Tricks?
Hello - Usually I have xscreensaver installed suid so that it can grab password infos - it is started from an ~/.xinitrc by my normal user - I'm in the process of updating packages and this no longer works (only entering the root password unlocks the screen now) Log messages for xscreensaver -verbose between the working 5.11 copy (old box - running 2.10) and the new box (xscreensaver 5.15 / df-2.13) are roughly identical Any ideas? Can probably dig around on this but pinging first in case anyone has run into the same issue of late not digging on this gives me time to dig on GConf stuff from earlier thread Cheers, - Chris
Re: repeated reboots
On Thu, Feb 02, 2012 at 12:47:57AM -0500, Pierre Abbat wrote: Also, when Kmail started, Akonadi complained about not being able to register with dbus. I thus have to write this on the Linux box. How do I fix it? I think this is roughly related to the GConf error you reported earlier that I was seeing as well - something seems amiss with gconf or dbus at the moment - I'm trying to write a test client to isolate the issue and if that works and the issue is not clear, I will try to see if there is some configuration/pebcac error, and if not, see if some gconf/dbus experts can assist determining if this is a bug there, or something on the df side For refs, building dbus w/o KQueue support did not seem to help the issue along for me, fwiw - and I enabled a system dbus and hald prior to both tests which I had not previously enabled. Cheers, - Chris
Re: Akonadi/dbus was: repeated reboots
On 02/02/12 05:31, Pierre Abbat wrote: I have 2011Q4 in /usr/pkgsrc, but Q3 in pkgin, and have rebuilt one package. Should I try building dbus in Q4 or wait till the packages come out? Mine is using Q4 (technically -HEAD during the Q4 branch period of a couple of weeks ago) - Further tests on my end (nautilus is my target application) seem to indicate there could be corruption in the GConf database triggering the error - or maybe mis-installation in the pkgsrc scripts - still not clear if this is the case or not as I'm about 50% through figguring out how to manually manage/debug the GConf DB to find/rule that out. Building the perl GConf bindings and hacking the example up to 'dbus add directory' is enough to trigger the 'cant connect to session bus' error on my system. A simple python/dbus 'connect to bus' program seems to properly connect however. I'll keep chipping away on it until I figure it out - might take a few days. After that I'll put up some notes on investigating these kinds of issues since there seems to be a dearth of 'installing/configuring/administering dbus/gconf' docs out there. Though, to note, the connection to the kmail and especially konversation issues was totally a guess based on the error initially reported about connecting to the session bus. Cheers, - Chris
Re: Paper size
On 02/02/12 14:00, Pierre Abbat wrote: paperconf outputs a4. Since both are set to A4, why do the apps insist on defaulting to Letter? Because it is the 1 true paper size :D j/k even though from the usa I prefer a4 anyhow ... Seriously- the thing about unix printer setups and configuration tools is that there are so many to choose from - some are going to use PAPERSIZE env var, some will use paperconf, some will have their own (eg. gnome-print), etc. Probably the printer is printing whatever raw data is given to it from the application - so its likely in the program config rather than printer setup config or one particular configuration tool. Don't have any specific tips for these applications - but I'd check the docs for the individual apps 1st. Cheers, - Chris
Re: OpenJDK 1.7 build notes
Update: On Tue, Jan 03, 2012 at 05:57:54PM -0600, Chris Turner wrote: x64 built through 1.6, however am encountering some build errors on 1.7. These do seem to be build configuration errors, and not e.g. JVM runtime errors, so I suspect they will not be too hard to correct. Got 1.7 to build properly on 64 bit from 1.6 without much tweaking - basically the 'defs.make' needs to understand x86_64 as well as amd64 - so for the impatient - follow the previous 1.4,1.5,1.6 bootstrap instructions + bmake configure hack in x86_64 block to bsd defs make by copying the amd64 one bmake install package should yield a 1.7 install package. Am currently doing a sanity rebuild to make sure I've recorded all of the changes, then will try a native 1.7 - 1.7 hosted build - If all is well there, will be working to get the patches / boostrap tarballs in after that. Cheers, - Chris
Re: ssh -Y doesn't work
On 01/16/12 11:21, Pierre Abbat wrote: Warning: No xauth data; using fake authentication data for X11 forwarding. ^ this is the problem Make sure that your DISPLAY is correct on this box, and that X clients can connect to your X server from here before trying ssh. Also always a good idea to make sure you *need* ssh -Y instead of plain-old ssh -X. Alternately you can likely muck with the X cookies on the remote end - see also xauth(1). Cheers, - Chris
Re: OpenJDK 1.7 build notes
On 12/18/11 02:21, Chris Turner wrote: I'll also try to repeat the process on x64 and see where that goes as soon as I get a chance to get a build / test environment setup (probably a few weeks) Update: x64 built through 1.6, however am encountering some build errors on 1.7. These do seem to be build configuration errors, and not e.g. JVM runtime errors, so I suspect they will not be too hard to correct. For those wanting any jdk on x64 - the makefiles previously announced along with your own copy of the JDK source should build the 1.6 system, with the exception that jdk 1.6 required a hack to the pkgsrc 'gcc' wrapper to -I the correct directory for freetype (workdir-include/freetype iirc), and for some reason the pkg_alternatives setup didn't work on my 1.6 package, so to use you might need to link these in manually, etc. However, I have run over my alloted time to proceed at the moment, so I will likely be picking this up again in a few weeks time. Cheers, - Chris
Re: What is the minimal memory requirement for HAMMER?
This value is tunable - an old P4/512Mb box required me to do the below locally, which alleviated the issue (though did impact running memory) See also the description in the LINT config / your current running values based on your warning looks like you'll want to try 4096 or so if your system can afford it. (not sure if this has to be ^2 but always good to assume things do when dealing with kernel knobs) Cheers, - Chris $ git diff GENERIC diff --git a/sys/config/GENERIC b/sys/config/GENERIC index e117741..7a1e790 100644 --- a/sys/config/GENERIC +++ b/sys/config/GENERIC @@ -4,6 +4,10 @@ # Check the LINT configuration file in sys/config, for an # exhaustive list of options. +# dev02 low memory / hammer cleanup hack +optionsNBUF=8192 +#end dev02 low memory / hammer cleanup hack + platform pc32 machinei386 machine_arch i386 On 12/20/11 11:39, Zenny wrote: Hi: I am trying to experiment with HAMMER with a spare PIII/500Mhz old machine with 256MB of RAM. I installed the RELEASE-2.10 version of DFBSD. But when I try to execute hammer cleanup, it spits out an error that reads: hammer: System has insufficient buffers to rebalance the tree. nbuf 3969 However, when I checked in the archive of the mailinglist (http://leaf.dragonflybsd.org/mailarchive/users/2010-03/msg00057.html), it states that 128MB is the minimal requirement, right?! In that case the hw that I am using exceeds the minimal requirement and still spits out the error, which reportedly due to the low memory (http://leaf.dragonflybsd.org/mailarchive/users/2011-05/msg00074.html) Please throw some light! (Very dark here in Scandinavia during winters like this) ;-)
Re: dragonfly bsd and vkernels
On 12/08/11 03:18, Samuel J. Greear wrote: Vkernels are mostly used for kernel development and testing, but also certainly for isolation. Side note, for reference, at risk of stating obvious - we also still have jails as well. Cheers, - Chris
Re: Problem booting
On 11/29/11 03:06, Konrad Neuwirth wrote: The only thing that I could reasonably easily produce is an output of lsusb --verbose from a debian console on that hardware, I'm enclosing the output as a txt file. The stuff that I am particularly interested in is the Raritan interfaces. Of course you were correct, there is a HID device there, a virtual mouse. But after that comes what I am interested in ? the virtual storage device. Hmm - on further inspectino looks like this is not verbose booting - could you try to get the dmesg from a verbose boot? also - not clear if it's possibly to attempt disabling the mouse on this device either - just in case there are interaction issues that would be a good thing to try as well. Cheers, - Chris
Re: Problem booting
On 11/22/11 04:13, Konrad Neuwirth wrote: The keyboard the devices has is well recognized; but after the ums0 probe, nothing happens. Theoretically, there should still be a CD Rom attachment on the device; that is what I'm trying to boot off. Is the ums port the CD device being mis-recognized or is this some kind of virtual mouse device? If it is a virtual mouse or similar, you might want to disable this device and retry - if it is the CD - you might wish to boot another OS that works and provide the usb and PCI device info (e.g. 'usbdevs -lv' / 'pciconf -lv' on dragonfly ) - perhaps some simple device ID's are missing somewhere that could be added. It might just be simpler to get someone to physically connect a USB key or USB Cdrom and try that instead for now - the other device items could be fixed more easily from the machine once it is installed. Cheers Good luck, - Chris
Re: my old laptop bios freezes on reboot after dflyBSD installed
On 11/22/11 20:42, Edward M. wrote: I'm have an issue with my old laptop and DragonFlyBSD. Does the bios have an option to disable ACPI? This might be worth a shot. You might also muck around with IRQ numbers for the various devices if that is possible. Also - you might want to make sure there are no errors when installing the bootsector on the HDD, or manually install (see the text file on the live cd root) and 'dd' the drive out before creating filesystems Cheers, - Chris
Re: Unable to boot Dragonfly GUI on virtual machine
On 11/09/11 15:00, Sanath Kumar wrote: PS: I would love to install DFLY on my laptop, but the X Server won't start because my laptop has a non-standard NVIdia hybrid gfx card. So, I am trying to install it in the VM. Not sure how nonstandard you mean, but usally the 'nv' driver works (without 3d) on most nvidia cards.
Re: partition invalid or corrupt (solved)
On 11/13/11 01:38, william opensource4you wrote: In fact Grub sounds to not be able to mount the UFS of Dragonfly ;-(. So, I must ask Grub to forward the control to an another bootloader. Wonder if grub only understands 32bit disklabels? hmm.
Re: Streamline pkgsrc issues: DragonFly developer gained NetBSD commit privilege
On 09/12/11 05:17, John Marino wrote: I think the dfly-pkg-people idea was probably okay in theory, but it doesn't sound like it's been too successful so far. Personally I think this is symptomatic of the fact that our overall df-to-pkgsrc bug reporting / fixing process could be much more clearly defined - e.g.: - df wiki page on the 'official' procedure to get something included (e.g. some of this might just point to regular pkgsrc procedure - and other stuff might relate to getting 'our' committers to pick something up, etc) maybe outlining this by 'class' of issue - e.g. new packages: should be submitted to pkgsrc-wip or what have you, whereas a 'branch bugfix' might be submitted to netbsd bugs and notify the df-people, etc. - some kind of cross-notification about what bugs are in the system / what needs fixing, etc. - perhaps some kind of information about who has commit access and/or a mailing list df-side to bang on people to get things 'properly' verified / cleaned up / tested / committed etc. No need to wait on some specific person for a one-liner ./configure script patch, etc. to make some package assigned to pkgsrc-users@ work, on the latest pkgsrc branch / df relaease, etc. The bulk build reports are great, as is the query pr stuff, and having pkgsrc committers - but we just need some 'glue' to bring it all together IMHO which would probably help improve things a ton Seems like a df-side mini-project which serves as an official liason with the pkgsrc/netbsd side people would fill the need here - not trying to usurp their processes by proxy but at the same time there are some aspects that are more-df-than-netbsd here and probably make sense to be handled as such. I'm happy to pitch in where possible if any of the above needs doing - just don't feel like I have the 'authority' to set the tone, policy etc - cheers, - Chris
Re: Seeing processors induvidually on an SMP system dmesg
On 08/23/11 07:32, Siju George wrote: Is this normal? Shouldn't the cores be listed separately? You have to use 'top -M' to see the per-cpu usage percentage instead of the overall average. Individual processes should still be listed as running on individual CPU's in the process display without this option, however. Cheers, - Chris
Re: PPTP VPN
On 08/19/11 18:00, Pierre Abbat wrote: I've looked all over the Web and tried things and nothing worked right. And I still don't understand what's going on. What does pppd do, and what does pptp do? Why are they separate? There is a package ssh-ip-tunnel-1.0nb1 = Simple VPN system using pppd over ssh. I can ssh into the box. Can I use this and get the same IP address that I'm supposed to get with pppd and pptp? Do I have to log in as root? Pierre Oh Man. This is a can of worms :D opening. carefully. anyone feel free to chime in and correct me where I am wrong :D Basically - VPN is a generic term for an encrypted network tunnel, with many possible technologies - simple SSH/SSL tunnels, IPSEC tunnels, PPTP, OpenVPN, and others. PPP is the 'point to point' protocol - which can be used for setting up a variety of network links, most often modems / ISDN / T1+ lines but also other various configurations such as GSM modems, wide-area fiber optic links, etc. In DragonFly we have 2x implementations of PPP - one, 'ppp' which runs as a userland tool on top of the tun/tap device, another, pppd which runs in kernel space. There is also a netgraph(3) implementation of PPP - I don't recall if this is separate from the 'pppd' kernel version or used to support it . The ppp one is a bit easier to setup/debug due to the nifty command shell it has, but is slower performing than pppd due to the user-kernel data copying requirements. The 'ppp' (userland) version of PPP is common across the BSD's and differs from a different 'ppp' (userland) version which was implemented for linux with different commands, configuration, etc - however, all speak 'PPP'. PPP is a serial communications protocol, and as such is often setup over terminal lines (such as a dialup modem tty, etc) - it appears that the ssh-ip-tunnel package uses the 'dial up tty' portion of SSH in combination with the '/usr/sbin/pppd' (kernel mode) package. Keeping in mind that the package is in pkgsrc and requires kernel features - the netbsd kernel mode pppd package has common ancestry with the netbsd kernel mode version and so the ssh-ip-tunnel might work for setting up these kinds of vpn's using the dragonfly kernel-mode pppd implementation. However, PPTP is a specific and somewhat standardized method / protocol of tunneling PPP (point to point) traffic over an existing IP infrastructure and as such you will need a PPTP implementation to connect to the remote end. As the pptp package is the client only one, this is the one you'll need to use - the freebsddiary post is a bit confusing because it defines the pptp configuration file as ppp.conf - however, whatever the file is called it needs to be called using the pptp program. Usually / generically speaking for VPN's in general there is some primary connection which is established to negotiate the configuration of the actual tunneled connection - the exact specifics of the negotiation and what authentication is required will vary according to the given setup / choice of VPN technology and other administrator settings - again not being versed in PPTP and your particular situation, I can't say what this might mean for your case, e.g. root, dialup, foo, etc. Hope this clears things up a bit / somewhat. Or makes them murkier in a way that might lead to clarity :D And keep up your hope - I think if one can configure a PPP or VPN link and a mail server 'blind', one can pretty much configure anything in the world of computers so this is 50% of the battle :D Cheers Good luck. - Chris
Re: PPTP VPN
On 08/20/11 05:53, Chris Turner wrote: As the pptp package is the client only one, this is the one you'll need to use - the freebsddiary post is a bit confusing because it defines the pptp configuration file as ppp.conf - however, whatever the file is called it needs to be called using the pptp program. ach - looks like I misread your output - you did call pptp which called pppd and had some errors. This could be a few factors - 1: pptp is compiled from pkgsrc - which might mean it is expecting a netbsd pppd underneath instead of the freebsd-dragonfly one we have (need to check freebsd4/freebsd-now/netbsd pppd manual pages / code) 2: pppd has drifted w/r/t pptp or the other way around and pkgsrc needs updating, (need to check pptp manual pages / code ) 3: ... To get this moving in the right direction, you might want to try setting up a spare freebsd box with ports or a netbsd box with pkgsrc, get the pptp link working, and try to figgure out the possible pppd and pptp differences on dragonfly that are keeping this from working - aka, you might need to port/update pppd or pptp, or pkgsrc, or 1/2/both/all, etc to the current state of affairs on DragonFly This is why I keep sources for all BSD's handy for troubleshooting purposes :D Alternately you might just want to try SSH port forwarding if you can SSH to the remote end :D if anyone has a working PPTP setup this would be an excellent time to chime in :D cheers again and good luck. - Chris
Re: rebuilding pkg_install fails
On 08/11/11 16:52, Pierre Abbat wrote: If the maintainers fix the bugs, will the fixes get into the git branch so that I can get KDE, even if I have to compile it? If the bugfixes don't make it into the stable quarterly you are using you can always do something like: mv /usr/pkgsrc/category/brokenpackage /usr/pkgsrc/category/brokenpackage.bak cp -r /usr/fixed-pkgsrc-tree/category/brokenpackage /usr/pkgsrc/category/brokenpackage for the various broken packages and use your 'hacked up' tree to build the related packages. This works most of the time unless there have been changes to the various dependencies of the broken packages, in which case you'll need to do the same for those, or tweak the 'fixed' packages to use different versions in the package makefiles. Cheers, - Chris
Re: hal and udev
On Sun, Jul 24, 2011 at 09:21:58AM +, Roelof Wobben wrote: Disclaimer: not a gnome dev / only familiar at high level - It is correct that Dragonfly uses udev instead of the deprecated hal. As I take it - hal is a gnome related hardware layer so e.g. gnome applications can do things like mounth thumbdrives, etc without needing to directly interface to the underling os - whereas udev is the actual linux devfs implementation - though I haven't dug into gnome lately so maybe theres a gnome udev as well these days.. or ?? dragonfly has its own devfs implementation - I'm not sure what the gnome VFS layer uses to interact with this .. And which version of Gnome is avaible ? This depends on which version of pkgsrc you are tracking - if you're wanting to use binary packages we typically follow the latest stable branch - http://pkgsrc.se/x11/gnome-desktop shows that pkgsrc-CURRENT is at 2.32 and the last few pkgsrc stable branches have been at 2.30+ The various makefiles used in the pkgsrc tree should give you an idea of the build options selected. You should also be able to browse the available binary packages for dragonfly though I don't have URL handy at present. Cheers, - Chris
Re: Running OpenGrok on DragonFly
On 07/23/11 04:32, Francois Tigeot wrote: My motivation for making the JDK 1.6 work on DragonFly was to run OpenGrok. This is a great document - could you add a 'howto' to the 'howto' section here: http://www.dragonflybsd.org/docs/documentation/ and maybe a separate page for just the java portion? similarly if anyone else has any 'how to' steps for some of our quirkier packages e.g. ooo, updates for flash, etc - maybe now is a good time to solicit update help? (probably should respond on docs@) cheers, - Chris
Re: Unable to login with PPPoE (2.10-RELEASE x86)
it that way), I just wanted to test it.: ---snip- root@aquina /etc/ppp cat ppp.linkup MYADDR: delete default add default HISADDR ---snip- Im going to attach my ppp.log again (renamed ppp_2.log). You'll notice a specific line: Jul 7 19:16:24 aquina ppp[337]: tun0: Warning: 84.63.128.1: Change route failed: errno: No such process I think that this is about the problem, but Im pretty unsure how to interpret that. cheers Georg Am Donnerstag, den 07.07.2011, 10:05 + schrieb Chris Turner: On Thu, Jul 07, 2011 at 10:50:46AM +0200, Georg Bege wrote: The funny thing is, addresses do get resolved (if I dont have any default) I dont get anything (no dns/resolving). But ping doesnt get through nor any kind of connection. Do I have this correct: - without the route assignment, dns lookups are working - with the route assignment, dns lookups are not working? I didn't actually see the address assignment in your ifconfig output - was the ppp0 device output truncated or ? If I am correct, it sounds like perhaps some data is flowing but not all - which could indicate an MTU mismatch Though I don't have direct pppoe experience on FreeBSD/DragonFly - I did once use the OpenBSD implementation (which uses a userspace daemon rather than the netgraph device) - I had some issues with the mtu matchup which caused some issues - My archived configuration had : set mtu max 1492 # set mru max 1492 so you might try commenting out the mru portion? What does the ping of e.g. the remote gateway, show exactly? host unreachable or something else? Its been a while since I debugged an MTU mismatch but iirc if you can ping the gateway but not something else (like say your upstream dns server ) and the routes look ok (and tcpdump is showing the packets flowing out ) you can set some combination of ping -p and -D to detect the mismatch again, IIRC, I think e.g.: ping -p 1500routed IP? gateway IP? ping -p 1492routed IP? gateway IP? - works ping -D -p 1492routed IP? gateway IP? - works ping -D -p 1500routed IP? gateway IP? - fails or something like this - I'd verify what I'm saying with some searching :D or maybe someone can chime in? that wouldn't explain why the config is working on one but not the other, though I do recall that some of the default settings might be different w/r/t freebsd, etc. in ppp from previous adventures with PPP (using GSM devices) good luck! Cheers, - Chris
Re: Unable to login with PPPoE (2.10-RELEASE x86)
On 07/08/11 11:15, Georg Bege wrote: Hi I've to say that I've got it to work finally! I was playing with it yesterday evening, and suddenly it work - Im unsure if it was a typo or whatever - really dont get it. But its okay now! Nice! I forgot to warn about the 'dreaded ppp.conf indenting' issue so maybe that was it :D care to paste the ppp.conf back so any errant googlers are satisfied in the future? cheers, - Chris
Re: desktop or server
On 07/07/11 09:45, Georg Bege wrote: Personally I'd say you could use it for both, but what you've to keep in mind are proprietary graphic card drivers (as nvidia). to be clear - these cards almost always work in console and for X in 2D mode with the xorg 'nv' driver. 3D support is not available cheers, - Chris
Re: Unable to login with PPPoE (2.10-RELEASE x86)
On Thu, Jul 07, 2011 at 10:50:46AM +0200, Georg Bege wrote: The funny thing is, addresses do get resolved (if I dont have any default) I dont get anything (no dns/resolving). But ping doesnt get through nor any kind of connection. Do I have this correct: - without the route assignment, dns lookups are working - with the route assignment, dns lookups are not working? I didn't actually see the address assignment in your ifconfig output - was the ppp0 device output truncated or ? If I am correct, it sounds like perhaps some data is flowing but not all - which could indicate an MTU mismatch Though I don't have direct pppoe experience on FreeBSD/DragonFly - I did once use the OpenBSD implementation (which uses a userspace daemon rather than the netgraph device) - I had some issues with the mtu matchup which caused some issues - My archived configuration had : set mtu max 1492 # set mru max 1492 so you might try commenting out the mru portion? What does the ping of e.g. the remote gateway, show exactly? host unreachable or something else? Its been a while since I debugged an MTU mismatch but iirc if you can ping the gateway but not something else (like say your upstream dns server ) and the routes look ok (and tcpdump is showing the packets flowing out ) you can set some combination of ping -p and -D to detect the mismatch again, IIRC, I think e.g.: ping -p 1500 routed IP? gateway IP? ping -p 1492 routed IP? gateway IP? - works ping -D -p 1492 routed IP? gateway IP? - works ping -D -p 1500 routed IP? gateway IP? - fails or something like this - I'd verify what I'm saying with some searching :D or maybe someone can chime in? that wouldn't explain why the config is working on one but not the other, though I do recall that some of the default settings might be different w/r/t freebsd, etc. in ppp from previous adventures with PPP (using GSM devices) good luck! Cheers, - Chris
Re: SORBS listing
On Fri, Jun 24, 2011 at 04:30:08PM -1000, Peter Avalos wrote: On Sat, Jun 25, 2011 at 12:15:32PM +1000, Dean Hamstead wrote: SORBS delisted crater, so hopefully that should help if anyone's using it. Cheers - That's what I get for being lazy and not hooking up http://www.policyd-weight.org/ to my mail setup, yet :D
SORBS listing
Was reading my 'lack of' mail - looks like somehow crater got blacklisted on SORBS - maybe someone signed up with a bad address? also saw a similar issue with freebsd lists so my tinfoil hat theory is to blame a nefarious GNUsurper with a vengance against 'non free' software - that, or their list is compromised. or they are two completely urelated incedents. Anyhow: http://www.sorbs.net/lookup.shtml?69.163.100.198 for someone 'authorized' for this kind of fixup Cheers, - Chris
Re: HEADS DOWN: via padlock possibly broken on master
On Sat, Jun 18, 2011 at 01:29:21AM +0100, Alex Hornung wrote: I added support to make use of Via's on-chip RNG, but someone needs to test it to confirm that it works as expected; I don't have a VIA CPU for testing. As someone recently bit by a similar type of issue ( rum(4) not being up-ported for new wifi stuff) this seems like a bad idea - If the code is untested and not 'core' -- why not commit it with some kind of 'options' guard or somesuch until someone is able to test, etc? /2c - Chris
Re: usb wifi stick
On Mon, Jun 06, 2011 at 08:28:44AM +0200, Sascha Wildner wrote: On Sat, 04 Jun 2011 13:26:34 +0200, Andrew Boehringer andrewboehringer...@hotmail.com wrote: Currently, we don't have a working USB wifi driver, unfortunately. What happened to rum(4) , for example? I do see it is in tree but not built in GENERIC - I could swear I used that in the ~2.6ish days - granted it is a 'g' device - but, this is relatively recent. Cheers, - Chris
Re: Intel Corporation 82578DC Gigabit NIC support
On Tue, May 31, 2011 at 12:52:15PM +0800, Sepherosa Ziehau wrote: You probably need to change dev/netif/igb instead of em and ig_hal. You could simply add the PCI ids to igb and see whether it works or not. The plot thickens - based on a whole lot of suppositions, b/c it would take hours to review the history and someone here probably knows it better anyhow - It looks something like em driver (perhaps btw v6-7 of the intel rev?) grew into the e1000 driver, with pcie support - and a 'legacy' em(4) driver - subsequent updates? were made to the e1000/*em* items, such as adding more chips - Our e1000/em contains what apppears to be the needed items for this card: em0@pci0:0:25:0:class=0x02 card=0x80001025 chip=0x10f08086 rev=0x06 hdr=0x00 e1000_api.c:case E1000_DEV_ID_PCH_D_HV_DC: e1000_hw.h:#define E1000_DEV_ID_PCH_D_HV_DC 0x10F0 if_em.c:{ 0x8086, E1000_DEV_ID_PCH_D_HV_DC, PCI_ANY_ID, PCI_ANY_ID, 0}, (the 0x10F0 being the relavent bits) which are not in the 'live' sys/dev/netif/em driver - however, this copy of 'em' (sys/dev/netif/e1000/em) is not linked into the build - whereas the sys/dev/netif/em copy is. I was able to : cd sys/dev/netif/e1000/em wmake to get a clean 'if_em.ko' - no idea if this will probe/attach/send packets and so on.. So - Siju - feel free to build that and test I suppose - not sure where to take it from here w/r/t 'em' development - either to disable / merge in the relavent bits of sys/dev/netif/em to e1000/em, or the other way around, etc.. my thoughts would be to use the new copy - however it does appear that someone at some point took the time to cleanly separate the phy stuff, into that 'ig_hal' etc- so maybe this was overlooked when e1000 was added, I dunno.. (see history, above :D ) so - I will defer to the experts cheers - Chris
Re: md5 sums and hammerfs encryption
On Wed, May 25, 2011 at 10:47:19AM -0400, Pierre Abbat wrote: Also, if root is not encrypted but some other partition is, can the init script time out and continue booting without the encrypted partition? For rebooting remotely this would be useful. Though I have not experimented with our LVM / encryption, etc - you might just set these partitions as 'noauto' in your /etc/fstab so that the boot is sucessful, and then write a quick little script you can run after boot up to configure mount the drives, etc.
Re: ad8: FAILURE - READ_DMA48 status=51READY,DSC,ERROR error=10NID_NOT_FOUND LBA=809594688
On Mon, May 30, 2011 at 10:54:26AM +0530, Siju George wrote: Also How do I map ad8 etc to their sernos? The answer i got from this list earlier was to ls /dev/ and /dev/serno and match according to the order found there Looks like you're on the right track - but not quite there - the main thing is to use the device minor to match things up - on my system, for example: # ls -l /dev/ad6s1a crw-r- 1 root operator 20, 0x00020030 May 29 21:51 /dev/ad6s1a # ls -l /dev/serno |grep 0x00020030 crw-r- 1 root operator 20, 0x00020030 May 29 21:51 6RY18WCV.s1a # grep 6RY.*s1a /etc/devfs.conf link serno/6RY18WCV.s1a dev02.root # ls -l /dev/ |grep 0x00020030 crw-r- 1 root operator 20, 0x00020030 May 29 21:51 ad6s1a crw-r- 1 root operator 20, 0x00020030 May 29 21:51 dev02.root So as can be seen from above, the 'ad6s1a' device is showing up with the device minor of 0x00020030 - which corresponds to '6RY18WCV.s1a' and also, in my case, as I have setup devfs aliases, /dev/dev02.root . Running 'natacontrol list' might also be helpful: # natacontrol list ATA channel 0: Master: no device present Slave: no device present ATA channel 1: Master: acd0 COMPAQ DVD-ROM LTD163/GQH3 ATA/ATAPI revision 0 Slave: acd1 CDD4851 CD-R/RW/C2.8 ATA/ATAPI revision 0 ATA channel 2: Master: no device present Slave: no device present ATA channel 3: Master: ad6 ST3250410AS/3.AAA Serial ATA v1.0 Slave: no device present Cheers, - Chris
Re: Intel Corporation 82578DC Gigabit NIC support
Hello - On Mon, May 30, 2011 at 12:10:10PM +0530, Siju George wrote: Thaks for your reply :-) Which files should I edit for this? I took a look at this w/r/t the FreeBSD / openbsd drivers to try and provide some hints, and it looks like this is a 'non trivial' case - that is to say, there is various device-specific code for the driver which looks to be required to make this particular chip work. Sorry to get the hopes up! (removes foot from mouth) That being said, it also doesn't seem like there is *too much* chip specific stuff, so it probably would not be too much of a port. As I gave bad advice previously, I will take a crack at this tomorrow. FYI - related files are: sys/dev/e1000 in the freebsd tree and sys/dev/netif/em sys/dev/netif/ig_hal along with: sys/bus/pci/pcidevs in ours. grepping around for 82578 shows what appears to be all the related missing pieces. Cheers, - Chris
Re: Mounting DVD disk in virtual kernel
On 04/17/11 06:18, vasily postnicov wrote: $ mount -t cd9660 /dev/vcd0 /mnt -t udf perhaps?
Re: SO_NOSIGPIPE
On 03/12/11 08:55, Sascha Wildner wrote: Same goes for NetBSD and OpenBSD. Someone obviously confuses FreeBSD with BSD there. But hey, that's still better than confusing portable with should compile on all Linux distros. :) What - so you're not running the latest *BSD 'distro? I for one am running the Dragonfly 'distro' of xBSD, as well as the NetBSD 'distro', and the OpenBSD 'distro' myself. If this doesn't work I suggest updating your xBSD 'distro' - there are links available at distrowatch.org AARRGGHH :D
Re: Intel Corporation 82578DC Gigabit NIC support
On 03/03/11 22:47, Siju George wrote: OpenBSD detects it as em0 but dragonfly does not detect it :-( Is there any thing I need to do like kernel configurations etc? Or should I wait till the driver is ported? knowing nothing about the driver itself - a quick check of our manual vs their manual and it does look like the same driver I'd suggest hacking pcidevs and seeing what happens when you build it.. it might just work (or not, depending) basically pciconf -lv will show the device hex stuff, and you can hack the general pcidevs file, and the driver stuff that matches it when that's all in place, the driver should try to attach, and either work, not work, or cause a panic :D after that well.. it's a matter of giving up or trying to see what the differences are of course if someone else has this hw working, this is all moot. cheers
Re: Home stretch on new network - if_bridge looking better
On 02/25/11 01:33, Matthew Dillon wrote: Most informative cheers!
Re: Home stretch on new network - if_bridge looking better
On 02/24/11 11:50, Matthew Dillon wrote: http://apollo-vc.backplane.com/DFlyMisc/bridge1.txt http://apollo-vc.backplane.com/DFlyMisc/bridge2.txt So - reading over this - is it correct that the setup is roughly like: - assign a local interface (lan0) to a network - add this network to the bridge - create openvpn 'bridged' mode tunnels - add these to the bridge so the L2 bridge / STP will 'map' according to the state of the ethernet bridging, which in turn relates to the openvpn tunnel state? Without diverging any security sensitive whatnot, Is the VPN tunnel created to the ISP or to say, the colo space? (I'd assume the latter) Have been working on my own openvpn (routing mode) fun to a pair of VPS's as well over the last few days so this is of interest :D also - I note in the bridge2.txt file you 'cd /usr/pkg/etc/openvpn' before running - is this so openvpn can find the config files? if so - to note, you can add a 'cd /path/to/configdir' within the config files.. also - assuming you have statics on both end of the tunnels - why did you choose openvpn ethernet bridging over say IP layer + ipsec? (or even openvpn 'routing' mode) with something like OSPF or similar and - do you have hw crypto cards on either endpoint? (my soekris 486 gets a little bogged down by the crypto, which is why I ask) ok enough questions ;) its definitely fun trying to convert consumer internet into a 'real connection' :D - Chris (from a gigabit LAN piggybacked on a sometimes 56k wifi link)
Re: Dragonfly network changes
On 02/18/11 00:53, Francois Tigeot wrote: Do they offer IPv6 ? man gif(4) MUHUAHAHAHAA - C
Re: Dragonfly network changes
On 02/18/11 03:59, Sergey V. Dyatko wrote: So please, next time wrote your jokes correctly. man mdoc(7)
Re: Dragonfly network changes
On 02/18/11 11:09, Ulrich Spörlein wrote: FTFY Within each index entry, the title of the writeup to which it refers is followed by the appropriate section number in parentheses. http://minnie.tuhs.org/cgi-bin/utree.pl?file=V1/man/manintro.txt UNIX PROGRAMMER'S MANUAL K. Thompson D. M. Ritchie November 3, 1971 Clearly, there is *NO* established convention regarding this notation. Thank you all for pointing out my errors. I will cease the inhalation of vaporized freebase cocaine immediately. :D - Chris
Re: installed Postfix, no periodic message
On 02/15/11 22:24, Pierre Abbat wrote: I installed Postfix, which provides a Sendmail compatibility program at /usr/pkg/sbin/sendmail. The Sendmail binary was at /usr/sbin/sendmail. The periodic jobs just ran, but I don't get the usual emails. How come? Do I have to add /usr/pkg/sbin to the path in /etc/crontab? I am as well reworking my mail setup to postfix - and have found the following most useful: http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/mail-changingmta.html original FreeBSD handbook http://www.dragonflybsd.org/docs/handbook/handbook-mail-changingmta/ our copy (haven't diverged too much) http://www.csua.berkeley.edu/~ranga/notes/freebsd_postfix.html some postfix specific stuff for freebsd - again, hasn't diverged too much http://www.ntecs.de/files/mailserver.pdf (kudos to M. Neumann for an excellent 'large scale mailsetup' using DF doc - good for pkgsrc specifics) Big #3 items for the switch are: - switch mailer.conf - enable / disable appropriate rc.conf services - enable / disable various /etc/periodic.conf jobs related to sendmail maintenance. (personally need to review these to see what's applicable what isn't / write replacements)
Re: installed Postfix, no periodic message
On 02/16/11 21:29, Justin C. Sherrill wrote: Whatever notes you make, please work them into the new handbook: for reference / posterity - most all is the same - Lots of documentation etc on my todo list this is definitely part - have been doing some heavy local-infrastructure rework for the past 8mo but as that draws to a close I'm very much hyped to begin putting my commit bit to good use :D ducks in a row and all that. cheers, - Chris
Re: rcrun reload
Pierre Abbat wrote: I modified a script so that it takes the reload argument, which the program supports, and then tried to use it: ... How come rcrun reload doesn't work? looks like rcrun itself doesn't support reload - usage: rcrun start|stop|restart|rcvar|list|forcestart|faststart|disable|enable I'd suggest using either the 'restart' target or maybe 'faststart' in your tor rc script to do the reload - alternately you could of course run: /etc/rc.d/tor reload directly as you demonstrated. the rc scripts themselves are shell scripts so the usual sh -x script args to trace execution, etc. should apply - for example: sh -x /sbin/rcrun reload tor to see the details of the particular error you saw main ones are /sbin/rcrun as you have found, /etc/rc, and /etc/rc.subr (common subroutines) of course the usual rc, rc.conf, rcorder, rcrun, etc. manual pages are a good place to look as well as of course the dragonfly handbook http://www.dragonflybsd.org/docs/handbook/handbook-configtuning-rcng/ although the associated 'rcrun' managment scripts are a dragonfly creation, The rc.d infrastructure originally comes from NetBSD - which has some good docs which are mostly applicable - http://www.netbsd.org/docs/guide/en/chap-rc.html there's also a PDF about the design as well as the evolution / history from which is quite excellent in the above URL ok. I stop the blabbing. As someone who has had to try and package SysV init S K softlinkk configurations, I get a bit exited about rcng and tend to ramble. cheers and good luck - Chris
Re: Comments on pkgsrc and DragonFly
On Fri, Jan 07, 2011 at 11:39:54PM -0500, Stephane Russell wrote: So at most, BSD forks can only be used seriously as strong servers. That's how I'm using dfly. FUD! Most of the things that don't work tend to be either: - obscure corners of 'thick' desktop environments (e.g. what do you mean I cant reencode video format APQYX from the media center plugin for qwertyfoo) - new, and primarily coded for linux and linux only - system interfacing things (e.g. interfacing with hardware at a low level) - os / memory / thread interfacing things (e.g. language interpreters that do funky things with memory / pointers / threads) General, traditional, classic 'Unix' stuff just works period. I'm using DragonFly almost exclusively - where I'm not is a requirement or due to wanting to have a machine for porting the stuff from the above. Use text files. Use traditional apps. Use an 'X workstation' and not a 'open source desktop'. Youll be fine. Yes - lots of user apps don't work. They're also fairly trivial to port. Its not an 'out of the box polished' kind of thing. But then - linux wasn't 5 years agao - and unix never has been. Neither is a notebook and a pencil. Still totally usable for end user apps.. just restructure the viewpoint/expectations. probably good for ppl psychically anyway /2c
Re: Comments on pkgsrc and DragonFly
On Fri, Jan 07, 2011 at 03:03:58PM -0500, Stephane Russell wrote: While porting programs to DragonFly, I had these issues (which are not bugs): Excellent thread - have you considered maybe making a 'porting software' page on the wiki? Thinking in general we could use a much more focused porting effort. ( I'm in the process of cleaning up a few pkg patches to this end meself) cheers - Chris
Re: Comments on pkgsrc and DragonFly
On Sat, Jan 08, 2011 at 03:11:57AM -0500, Chris Turner wrote: Excellent thread - yes - I realize this totally contradicts with my last post. I would like everything to just work - point is that will never happen unless ppl take the plunge and fix the bugs so - like camping - take whats necessary, dive in, etc /2c
Re: bug in Python wxWidgets
Pierre Abbat wrote: I'm running 2010Q2 on DFly. He suggested that I file a bug report, but I'm not sure whether the bug is in DFly or in pkgsrc. The versions are python26-2.6.5nb1 and py26-wxWidgets-2.8.10.1nb3. (The Linux box has Python 2.5.) The machine I have available for testing this is also a bit out of date at the moment - py26-wxWidgets-2.8.10.1nb2 python26-2.6.4nb4 - I got the error as well (i386 if that turns out to be related). I'd think the next thing to try would be to try a build against python 2.5 (or maybe even 2.4)- (see description for PYTHON_DEFAULT_VERSION in /usr/pkgsrc/lang/python/pyversion.mk) often there are issues with language changes and add-on modules with python version changes. First thing to do is to find the issue directly and notify the wxwidgets people - if they have a fix or a workaround it might make sense for a pkgsrc patch - but, possibly not (depends on their approach / where the bug is) probably a good idea to report you also saw the issue as well- that shows it's more widespread - it didn't look like from that thread that anyone dug too deep on it.. it could also be related to the swig version too - not clear on this. but yeah - first i'd try other PYTHON_VERSIONS / ensure the pkg's are using the latest versions. (see also pkgsrc guide ch20 , 21) and remember - porting the software is the *fun part* !! :) - Chris
Re: Can't connect to ssh
Francisco Reyes wrote: It is already set to bridged. Will look into VMware then if there isn't a built-in firewall or anything like it. Did the 'host' OS networking change? There are issues trying to bridge VM ethernet cards across wifi host adapters as wifi is not-exactly ethernet. What has me puzzled is that I have not changed the VMware config for the VM and it worked until the crash. Assuming you have access to the VM console (sounds like) - try running /usr/sbin/sshd -D -ddd on the server side and ssh -v on the client side and see if that shows anything obvious.. also maybe check the host firewall?
Re: lsof fails to configure
Pierre Abbat wrote: I'm trying to install lsof so that arm can watch connections on my Tor relay. ... How can I fix it? Sounds like it hasn't been ported yet - (anyone with news to the contrary?) typically this entails: bmake patch scratch around in work/ until build works ... where 'scratch around' = apply whatever OS-deciding logic detects freebsd to detect dragonfly, and modify if needed I have no idea about 'arm' internals - it might make sense to dig around there to see if it can use another interface to get the data
Re: Random x86-64 seg-fault finally fixed
Matthew Dillon wrote: Partitioning is already desireable for the current 48-core monster and I'd like to have some sort of DragonFly host guest solution that runs at full performance on the bare HW without virtualization. How would this be different than jail(8)? not understanding the 'without virtualization' part - I know some form of HW virt a-la kvm has been discussed a few times - do you mean like segmenting the 'machine' or somesuch?
Re: Encrypted root questions
Tim Darby wrote: Chris: agreed, email is not the best way to document things. I've experienced that same frustration trying to track down some important detail that only ever appeared in an email thread. well more my point was that DF stuff should be centralized on project infrastructure - but anyhow - cheers! - Chris
Re: Tor and Polipo
Pierre Abbat wrote: I'd like to start browsing the onion web. I set up a Tor node, which was pretty easy. I installed the Tor package, made a few changes to the torrc, forwarded a port, and was up and running. Firefox installed Torbutton by just clicking on a link. The missing part is Polipo. I can download and install it from source, but there is no package in pkgin or pkgsrc. How come Tor is in pkgsrc but not Polipo? my guess would be because noone has simply done it! there is squid - although it looks like polipo might be easier to configure - unless theres some other feature I'm overlooking?
Re: large ncpus / memory support going in, HEADS UP - master may have some instability
Matthew Dillon wrote: As part of the 48-core support some significant scheduler changes have gone into master, so again HEADS UP there may be some instability. How do these scheduler changes relate to say, userland preemption or things of that nature (thinking soft-realtime, scheduling priority, etc) ? I know in the past in general you've said thats not really a design goal - but, anyhow, seems to relate based on the commit messages cheers - Chris
Re: Encrypted root questions
Alex Hornung wrote: For whatever it's worth, I've added a task to google code-in a few weeks ago to document all this dm stuff, both cryptsetup and lvm, basically. A bit OT but shouldn't this stuff go in bugs / the wiki and then be referenced to any google code-in or foo barbaz-quux to prevent fragmentation? (from someone who has had to do 'where did that note go' many a time :) cheers
Re: less at end of file
Jeremy C. Reed wrote: less is not more. man uses more by default. you can: export PAGER=less or setenv PAGER less to change the man(1) behavior, btw.
Re: Firefox, Namoroka, Iceweasel
Ed Berger wrote: If you go into /usr/pkgsrc/www/firefox and do bmake show-options you'll see the trademark related mozilla-branding options that need to be explicitly set in mk.conf to build it as firefox. Alternately there are lots of user-agent modifying plugins for Firefox that should do the trick as well.
Re: Printer Daemon (Update)
Tim Darby wrote: Anyone have any insights into this? What FS type is your /var/spool ? I seem to recall some kind of tmpfs permissions issues of late.. cheers
Re: vlc assertion: z-z_Magic == ZALLOC_SLAB_MAGIC in _slabfree
Venkatesh Srinivas wrote: Do you happen to have the core dump? (It should be a file like vlc.core or something) in the directory where you ran VLC. also - as far as functionality goes in the meantime, many of these multimedia apps have selectable output plugins - you might have better luck using a different one for the time being? cheers, - Chris
Re: MC not starting
Paul Onyschuk wrote: I give up after fighting with pkgsrc for 2 hours. pkgsrc does a pretty good job of replacing the environment with a 'clean' one - so I'm not sure if the evnironment stuff would take effect - you can e.g. 'bmake -d A ...' to get more information than you could ever want about things.. also theres a PKG_DEBUG_LEVEL or similar to see the details - see also www.netbsd.org/docs/pkgsrc/ for sure perhaps just for tests it might make sense to move your system curses out of the way just to see what happens.. although, of course, ideally the 'proper' build should work.. I'm not well versed enough in the 'prefer pkgsrc' stuff to know how to debug it - I'd think: /usr/pkgsrc/mk/curses.buildlink3.mk /usr/pkgsrc/mk/curses.builtin.mk and maybe: /usr/pkgsrc/mk/defaults/options.description /usr/pkgsrc/mk/bsd.options.mk /usr/pkgsrc/mk/pkg-build-options.mk might be some things to check.. perhaps there's some bug there? another idea maybe is to try another curses-using package to see if it would build against your version - if so- perhaps the particular package makefile needs some kind of option set..
WRKOBJDIR (was Re: Hammer filesystem)
Matthew Dillon wrote: I agree that it should be setup that way on default installs. I don't know why NetBSD defaults to wanting to put the work directories right smack in the middle of a pkgsrc source tree. personally I'm agnostic here - I'll have a custom build setup either way - however: I could see a few reasons for this - 1) pkgsrc 'stuff' stays on the same partition 2) makes it easier to port stuff - as you just cd up or down a few ./..'s - e.g. diff file file ../../patches/patch-q cp hackedonfile ../../ for safe keeping, etc although, a few shell bits in ~/.profile sort that out.. 3) less confusing for new folks definately understand the desire to keep all 'build' stuff in /usr/obj too.. plus it's 'how things have always been done' (TM) right? ok. yeah.
Re: git: nrelease - gui - Change window manager, cpdup additional directories
Tron wrote: package available. But if die hard devotees of efficiency over glitz want Xfce - I am easy. don't you mean But if die hard devotees of glitz over efficiency want Xfce :)
SCons Pkgsrc
Kludging around in my local tree - Anyone with knowledge of SCons - or perhaps scons+pkgsrc - have any ideas why the below might be required? otherwise it expands to cc ... -W a l l ... other flags appeared ok - hence ellipsed in any event - it fixes the csound build on my Q1 branch haven't tested the program, but anyhow .. probably not worth a submit to pkgsrc and or upstream until I figgure it out.. I'm on py 2.4, if that matters.. Cheers! - Chris # cat local-patch --- /usr/pkgsrc/audio/csound5/work/Csound5.01/SConstruct.orig 2010-10-24 08:24:28 + +++ /usr/pkgsrc/audio/csound5/work/Csound5.01/SConstruct 2010-10-24 08:26:31 + @@ -284,8 +284,8 @@ # Define different build environments for different types of targets. -if commonEnvironment['MSVC'] == '0': -commonEnvironment.Prepend(CCFLAGS = -Wall) +#if commonEnvironment['MSVC'] == '0': +# commonEnvironment.Prepend(CCFLAGS = -Wall) if getPlatform() == 'linux': commonEnvironment.Append(CCF
Re: No package installation method works
Torbjorn Granlund wrote: It would be neater to have /usr/Makefile identify make and write a message: .INIT: @if [ `basename $(MAKE)` != bmake ]; then echo Use bmake stupid; exit 1;fi problem with this is that we *use* make.. for the system build. although - maybe it does make sense to have the system makefiles print a warning if called under a standard 'pkgsrc' path like /usr/pkgsrc, for these kinds of 'new user cases' it is pretty easy to get use to, imho.. but anyhow..
Re: sound no longer works for some programs
Pierre Abbat wrote: I'm installing it with pkgin, and I get this message: NOTE: Unfortunately, JACK wants to use a linux /proc filesystem... It's a DFly binary; does it know to look in the emulated /proc? ha - I think I didn't see this because I just built everything in /usr/pkgsrc/audio to see what would build.. In any event, you've inspired me to restart my testing anew - I'm currently running the following script to start jack (see below - non realtime, as regular user) and indeed, it does seem like I'm able to get some stuff from snd into the speakers (snd_hda) - and also routed through jack-rack and a simple 'echo' delay - (echo 'todo: port more plugins' todolist) I had to tune up my shared memory tunables to make this work - this needs to be done on a reboot - 2 entries locally are: kern.ipc.shmall=524288 kern.ipc.shmseg=65536 (which is also to accommodate pgsql - so YMMV) There's some hefty discussion history on the linux audio list about kernel schedulers, thread preemptability, etc in terms of ensuring that audio latency is 'suitable' for real-time use (e.g. playing your $instrument in time with the audio software without having to adjust for the delay) this in fact is a big part of what led to the soft-realtime scheduling misc. in the 2.6 linux kernel scheduler - and also IIRC wouldn't be possible to do on DF without heavy modifications to system calls, etc, etc, etc - I'm ignoring this as for now, as I'm mostly interested in sequencing, don't have nearly enough *working* to tax the processor, and also - all of this stuff might be less of an issue with multi-multicore processors (which weren't at all common when that stuff was going on) - as processor contention then becomes less of an issue. Anyhow - welcome to the beyond-bleeding-edge, or something.. - Chris --8-- $ cat dojack #! /bin/sh echo 'starting jackd' # check for shmget errors - # might need to ipcrm # fixme ipc key not 100% verified, but has been same across a few restarts jack_shm_key=2631977; ipcrm -M $jack_shm_key; exec jackd -d oss -r 44100
Re: sound no longer works for some programs
Pierre Abbat wrote: It's a DFly binary; does it know to look in the emulated /proc? to directly answer your question - no idea! but its doing something! :) - Chris
Re: sound no longer works for some programs
Chris Turner wrote: There's some interesting stuff going on in OpenBSD w/r/t midi - Based on a check of the NetBSD manual (and not the source) - it appears that NetBSD has grown a divergent (w/r/t OpenBSD) midi(4) as well..
Re: sound no longer works for some programs
Pierre Abbat wrote: What's jackd? Jack is a sound server / time transport sync patching setup designed mainly for audio production / music / etc - originally designed for linux but has since been made portable: http://jackaudio.org/ It's nearly-OT but there's quite a bit about multimedia production / music for linux here http://linuxaudio.org/ btw - theoretically alot of this should work if the audio + midi layer can be made to work - and some of it already probably does a severly time-deprived side project of mine is trying to get as much of this stuff tested or working on DragonFly as I can. So far, I haven't done anything really - jack runs. I've gotten snd (https://ccrma.stanford.edu/software/snd/) to talk to it. although I didn't have enough SHM configured for jack, and it therefore didn't make sound - and dynamically setting SHM freaked out my dev box - I've subsequently been too involved with web stuff and to risk the crash.. (see time-deprived) Is MIDI I/O supported in the kernel? The driver is snd_ich; does that kind of card support MIDI? no idea on this stuff - on my time-deprived todo list. There's some interesting stuff going on in OpenBSD w/r/t midi - see http://www.openbsd.org/faq/faq13.html#midi for an entry point - I think (haven't followed closely) thats mainly being driven by: http://www.caoua.org/midish/ also, according to my latest pkgsrc builds, some of the linux 'alsa' layer has been made user-space, so theoretically some software-side stuff might work. I'm overdue on some wiki edits - will retest type something up on what I have when I do that perhaps anyone else interested in audio/media production on DF should sound off here so we know who to pester :) cheers - Chris
Re: Is Citrix client working on dfly?
Tomas Bodzar wrote: However it was still not running so I dived in to the script and found that there is test for OS. This test is made by 'uname -s' and case for BSD systems is ...*BSD), but DragonflyBSD shows DragonFly. So I modified it directly in script and after that installation went fine and without problems. I will test real funcionality tomorrow, but I It's really amazing how much stuff this fixes, for the record - if anyone is thinking porting stuff to dragonfly is always tough - its not (always, that is) definately worth a 2 minute grep through any broken builds for something like 'FreeBSD' (NetBSD, OpenBSD, etc, etc) before giving up on making software 'X' work on DragonFly little programming knowledge required /2c
Re: Suggestion for hammer cleanup
elekktrett...@exemail.com.au wrote: Suggestions? quick-fix / hack wise - probably setup some job to run way more often that checks the status makes a determination - or move the job to something like anacron, etc although, in a laptop situation - you might want to manage this manually - because the heavy work from reblocking, etc could easily suck the juice out of your battery when you dont want it..
Re: sound no longer works for some programs
Pierre Abbat wrote: I open a sound file in Wavesurfer and attempt to play it and get silence. It used to work months ago. I try to play a MIDI file in Timidity and it says Couldn't open output device. XMMS still works, as does catting to /dev/dsp. Any idea what's wrong? I'm not sure what the sound card is, but there is no /dev/sequencer (which means KMid doesn't work). Pierre no idea about this particular app - but I do know lots of things are gradually moving to jackd and/or pulse audio, (maybe) with fall back support for traditional OSS/(alsa in the linux case) - Have you run it from the command line / checked any arguments there? I wasn't able to tell much from my build - it looks like it uses: http://www.speech.kth.se/snack/ for the sound routines - that page shows an example that appears to run (but not work, with no audio errors) on my i386/2.6 system as below - Perhaps a dig through the snack sources are in order? -- 8 -- $ cat /tmp/tst.tcl #! /usr/pkg/bin/tclsh package require snack snack::sound snd snd read /path/to/some.wav snd play
Re: Suggestion for hammer cleanup
Samuel J. Greear wrote: That said, I think it would be fine to commit one or more optional stopgap measures/scripts to the RC system, for mobile users and etc., as long as it is well documented that they may go away if a better solution is developed or derived. not to flamebait or something - but: wouldn't the system crontab be a better place? or perhaps create an /etc/periodic/hourly ? the 'newsyslog' is in /etc/crontab which is the only hourly job I can think of off hand.. the script can still of course source rc.conf or something to configure the job if the job is enabled by default
Re: MP BIOS bug on qemu from pkgsrc-2010Q2 on i386? - no linux boots
Siju George wrote: MP BIOS bug: 8354 not connected to IO-APIC this sounds like more of a linux question so probably better to check around in linux areas if I'm wrong - but you might try adding 'noapic' or maybe it's 'apic=off' to the 'kernel=' line in the bootloader - should be 'e' in most grubs, then 'e' again on the kernel line, then 'b'? to boot your edits for testing another similar idea might be 'noacpi' just to see if that changes it, perhaps change up your emulated hardware last I tried I was able to boot but still had some timer problems (like 'sleep 1' taking 5s) on linux that I couldn't quite figgure out - but this seemed to be more to do with qemu's timer emulation and/or newer linux kernel interactions or something.. anyhow.. right. good luck
Re: Linux Emulation Docs
Siju George wrote: I got flash working. Sites like youtube google videos work. Though I haven't compiled in the pcm module for sound good news - this means the doc is reproducable yay. will make updates tomorrow. are you on 2.6 or development? you don't get flash streaming video. refresh it a few times and X will freeze hmm. I think I got this once - but wasn't sure if it was the plugin or some interaction with the highly experimental windowmanager I was running at the time :) definately got some stuck npviewers at some point - and a lockup or 2.. actually disabled my plugin today after a freeze - but good to know it mostly-works 'in case' it's needed for some site or another I think perhaps something isn't quite 100% with the linuxulator threading / df threading stuff - Although that is entirely a theory - haven't had a chance to dig any further - am trying to use the system for Real Work(TM) although not sure if it's worth digging into on 2.6 with all the lock related / MPsafe stuff going on in 2.7 - Should i upload it to leaf? I also took a core dump at that point if you want I can upload it also to leaf. I probably won't get a chance to dig into it - but perhaps other linuxulator-minded folks might be interested. typical 'binary blob' disclaimers apply, that being said, I don't think it should ideally lock up anything, other than perhaps firefox, imho. How Do i try the flashplayer 10? I think there is a .tar.gz or similar available from adobe - if you backup the existing ns-flash package copy in the .so to the same location, it might theoretically be functional with nspluginwrapper (/usr/pkg/lib/netscape/plugins/libflashplayer.so is location for me) sorry if I made that part sound 'easier' - my memory was a bit fuzzy on that point at the time.
Re: Linux Emulation Docs
Siju George wrote: If some one can point me to docs on Linux emulation setup it would be great! http://www.dragonflybsd.org/docs/howtos/HowToFirefoxandFlashplayer/linuxemu.html is not working. Hello - See the discussion thread on the docs@ list from 9/2010 entitled Flash-related docs on the site (created against 2.6 / i386 / ff 3.6 / pkgsrc Q1 / flash9) I apologize for not having made the updates yet. also if you want to give flash 10 a shot - not sure what the state of the art there is - would be good to have an update for sure I am on it in the next couple days for certain - please post back any findings to that thread and I'll fold them in. Cheers Good Luck
Re: firefox instability may be fixed now in HEAD
Pierre Abbat wrote: I think that when I last upgraded all my packages, it was to Q1. For me, a package upgrade is a big event, because I don't have much free time, and I have to see if the upgrade broke anything (I frequently find that I have to symlink libraries, because the upgrade doesn't replace libraries that have the same version number but are now linked against newer other libraries). I *highly* recommend a jail or a simple chroot for this, if you're not using such a procedure already. Makes updates *much* less disruptive - as you can test new packages / fix any problems before going live - similarly, you can keep the old copy around in case of problems.. only downside is the disk space.. but these days - not so much of an issue in most cases cheers - Chris
Re: firefox instability may be fixed now in HEAD
Pierre Abbat wrote: I have a 36 GB disk (actually 38 GB but 2 GB is swap), usually 45-55% full. 4.4 GB is in /home. Do I have enough room to try this? Hard to say - you'll need enough space to store: 1) base system (~300-500MB) 2) pkgtools (not sure) 3) any packages (so FF + dependencies, or what have you) 4) if building in the chroot or jail, enough space for pkgsrc, distfiles, $WRKDIR, etc. Worst case is your disk fills and you have to wipe the setup - although - where's the other 45% of space? might make more sense to use an area there with more available if possible - or create a new partition for scratch work if the space isn't allocated. I usually setup a chroot, backup any important user files, and then 'mount null' my /home so that my data is in synch between the two (its a little annoying to copy files inout) DISPLAY=:0 is your friend in these cases if using a mixed old new setup - e.g: chroot /path/to/chroot DISPLAY=:0 gui-app-under-test the instructions in jail(8) for building jails work for chroots as well - just chroot /path to enter instead of starting the jail.. you might need to mount devfs as well - I'm not sure if this is needed - dont recall off hand also - some things like gnome apps, etc. might not have their 'proper' environment (e.g. dbus, etc) to display fonts, etc. properly - keep that in mind if testing. Cheers Good Luck
Re: USB image
Tron wrote: Yes, I can boot from cd but would like to try for the big image (if it was available). aah.. silly me - thought it was.. there is a 2.4 version if you're interested in that So if I understand you correctly, once the big image (with x, GUI, etc) would become available, I would: -burn it onto a flash drive yes or cd -change BIOS settings on the laptop to be the network installer yes, if by this you mean 'make it boot from the install media' ... -try out DF on the test box and if I like it install on HD from the network yes. although - I'm not sure how 'live' the 'gui live cd' is - perhaps others can comment? Another option of course is to boot up an emulator like VirtualBox or VMware, etc if you're just interesting in a bit of tinkering and that is an option for you - also I've heard that people are running under Xen/HVM recently although I'm not sure what host environment / Xen version that was.. (and hoping I'm not confusing things from the FreeBSD Xen list :) )
Re: USB image
Tron wrote: USB. (It is an old 700MHz Celeron with a BIOS that cannot be upgraded or easily patched because the mobo is oem ie: unknown...) you should be able to boot from cd, no? this machine is *far* faster than my old trusty 'bigred' - a spray-painted 266mhz amd k6-II haven't booted that guy in a while - but he was alive kicking in the 1.10 days.. or are you trying for the 'big package' image? can't remember the sizes off hand.. also: another option is to use the built-in network installer if the bios supports network booting has a nic basically, plug 2 machines into a LAN (if you have a hub or can borrow one), boot up one select it to be a network installer, and boot up the other one from the network side note - don't try hammer on a small drive - iirc 40G is the bare minimum for a light-workstation kind of setup? (remember you need to have space for file history as well as the files) no idea how this all works with the 'big package' version - perhaps others can comment.. cheers
Re: USB image
Chris Turner wrote: basically, plug 2 machines into a LAN (if you have a hub or can borrow one), boot up one select it to be a network installer, switch / crossover cable / etc should work fine too for sure - just referring to least-common-denominator
Re: ns-flash on df
Chris Turner wrote: to anyone interested I've posted some updated flash9 docs on the docs@ list- if the current perception is that it isn't working (which was my thought) - it is, for me at least. please note - not so stable. has caused a couple X lockups or - maybe that's interactions w/ my obscure WM. not sure. anyhow.. yeah.
ns-flash on df
sorry to cross post for anyone subscribed to both to anyone interested I've posted some updated flash9 docs on the docs@ list- if the current perception is that it isn't working (which was my thought) - it is, for me at least. cheers
Re: pkgsrc package builds for 2.8
Justin C. Sherrill wrote: The next release of pkgsrc, 2010Q3, is due out Oct. 1st. DragonFly 2.8 is going to be out soon after. I stopped the automatic builds of pkgsrc in the various places I'm building it, as I don't think there's going to be any changes to really catch at this point. whats the status w/r/t pkgsrc cvs, git our git mirror these days? last I recall there were some sync problems and the 'official' source fell back to cvs? side-note/theory thinking some sort of project-local branch might make sense as a way to funnel patches back to 'official' pkgsrc for inclusion.. nifty thing about distributed VCS is - well - it's distributed ! blah blah blah.. one day at a time cheers - Chris
Re: about dma.conf
Matthias Schmidt wrote: No, dma replaces sendmail/postfix (in parts). Sending through sendmail means that your MUA uses the local installed MTA (here dma) and not a remote smarthost. This is a known phrase. for reference as in: # /usr/sbin/sendmail -t EOF To: u...@host From: m...@here Subject: IT IS AN EMAIL! Hello! This email will be sent using the 'sendmail' binary to send mail through the local MTA! Doing email like this is very handy for sending myself reminders and script output too. I Like it! EOF rather than some kind of SMTP API in whatever program whatever application is written in..
Re: How to start CUPS?
Pierre Abbat wrote: I also see that GROFF_PAPER_SIZE is set to letter. Wouldn't it make more sense to set it to A4, which is the most common paper size in the world? BAH! Next think you know you'll want 4 holes instead of 3! /USA :)
Re: OpenOffice
Alex Hornung wrote: For whatever it's worth, I recently (~2 weeks ago) installed openoffice from /usr/pkgsrc (openoffice3-bin, iirc) and it worked just fine. What pkgsrc branch did you install from? Are you using the regular linux compat env? I'm still on Q1 - am getting the below Will be kicking off another build this week, so perhaps this isn't worthwhile ATM.. anyhow.. yeah. Cheers -- 8 -- $ grep 'Linux compat' /tmp/pkg_info suse_base-10.0nb5 Linux compatibility package suse_expat-10.0nb2 Linux compatibility package for expat suse_freetype2-10.0nb5 Linux compatibility package for freetype-2.x suse_fontconfig-10.0nb6 Linux compatibility package for fontconfig suse_libjpeg-10.0nb2 Linux compatibility package for JPEG suse_libpng-10.0nb4 Linux compatibility package for PNG suse_libtiff-10.0nb4 Linux compatibility package for TIFF suse_x11-10.0nb4Linux compatibility package for X11 suse_gtk2-10.0nb4 Linux compatibility package for GTK+-2.x sgi-fonts-1.0.457nb3 Linux compatibility package for SGI fonts suse_compat-10.0nb3 Linux compatibility package with old shared libraries $ ls -al /compat total 4 drwxr-xr-x 2 root wheel 512 Mar 18 2009 . drwxr-xr-x 20 root wheel 512 May 19 00:17 .. lrwxr-xr-x 1 root wheel 19 Mar 18 2009 linux - /usr/pkg/emul/linux $ ls /usr/pkg/emul/linux bin dev etc lib opt procrootsbinusr var $ swriter sh: /dev/null: Permission denied javaldx: Could not find a Java Runtime Environment! /usr/pkg/opt/openoffice.org3/program/soffice.bin: error while loading shared libraries: libXext.so.6: cannot open shared object file: No such file or directory