The Intersect

I finally finished grinding standing with an R&D corp so now I’m free to pursue missions with another corp that I have specifically chosen for their political allegiances and an item in their LP store with a nice LP to ISK conversion rate. Before I embarked on the journey to start my career with this new corp I decided to stop by Nourvukaiken where I had a few cheap ships stored that I never managed to lose in Tama.

I started with a Kestrel. Jumped in to Tama and began going through the belts. On the bottom belt I was tackled by a Crow and I lost my Kestrel without even putting up a much of a fight.

After this defeat I wanted revenge, so I went and got one my Caracal’s and jumped back in. Headed back to the belt and… nothing. He was gone. I went back through the belts, finding nothing at each one. So I warped to the gate and waited but gates are rarely exciting solo because 1) I can’t tank the gate guns and 2) anything I could freely shoot can’t approach the gate. So I warped to the station and got much the same result as at the gate.

After a while I went through the belts again which yielded nothing and after resolving to lose the Caracal I sat at the warp-in point on a belt and waited. After 5 minutes a solid-red (-9.9 sec) Kestrel showed up.

I suck at PvP so I was apprehensive about how this would go, but he was right in range so I scrambled him and fired my Caldari Navy Bloodclaws. To my surprise his shields were melting while I was taking almost no damage. About 30 seconds in a rat spawned, a Guristas Destructor, and to my relief went after the Kestrel. He exploded right after he hit armour, I thought he had self-destructed but I got a killmail so I guess not. And strangely, the guy in the Crow that had killed me earlier was on it. I guess he had been attacked by him and managed to escape before I killed him without him having changed systems. (He had taken no damage when engaged).

So there it is, my first real PvP kill on my new character (I had a few faction warfare blob kills before this). If you could count me being in a Cruiser, him in a frigate and him being attacked by a rat as ‘real’

Back to carebearing in my new mission home.

Last time I was questing my morality as an innocent corporation owner’s plea for surrender was cut off mid-sentence as his headquarters exploded. Today I’m questioning my mentality.

Seems I got a little caught up in purging the earth of democracy loving Gallente at the behest of my hyper-capitalist overlords and my standings with the Gallente are now -4.42. If they drop below -5 they will democratically elect to kill me whenever I enter their space.

Basically I got impatient. If the agent gave me a mission to kill Gallente I would decline it but if I got offered another anti-democracy killing-spree before the 4 hours were up I would invariably accept it, rather than waiting out the 4-hour timer.

I haven’t even started training Diplomacy yet so I have an escape route if things go too wrong, but I will have to do something about it soon. I really dislike having the areas of high-sec I can visit limited.

[15:44:43] Message > Gallente Company HQ: Unidentified ship, cease your attack! We are operating fully within both Caldari and Gallente law.

[15:45:09] Message > Gallente Company HQ: Unidentified ship, stand down! We are a civilian entity. We pose no military threat. We have no boosters, minerals, tech, or anything else of value. Please, stand down!

[15:45:47] Message > Gallente Company HQ: For the love of god, stand d–

So I am grinding standing with the Caldari Navy with my Industry alt to get rid of the refining tax. I got enough standing to move up an agent, the agent appears under the “available to you” list in the station but when I talk to them … nothing. They tell me to get back in line and talk to an underling.

’tis a bug. Log out and log back in and they will happily give you a mission.

I have a file lying around with instructions for getting around some OpenSolaris issues I’ve ran into lately, they may prove useful to others.

ld: fatal: file /opt/sunstudio12.1/prod/lib/crti.o: section .rela.annotate has invalid type [ SHT_RELA ]
ld is trying to do 64-bit linking on a non-64 bit system, do export LD_NOEXEC_64=1 and retry the compile.

readline warning turning off output flushing
This is a bug triggered by an upgrade from the 2009.06 stable to certain dev versions. Add clone:ptmx 0666 root sys to /etc/minor_perm

Broken keyboard layout, SHIFT+2 produces ” and SHIFT+’ produces @ (among other oddities)
This is most probably Acer Extensa 5620Z specific but pressing Fn+I fixes it. (Left size of Extensa keyboard: [CTRL][Fn][WIN][ALT])

More a tip than a workaround:

Update system
2009.06 has some very outdated packages and some important packages are not available in its /release repo. Complete upgrade:
pfexec pkg set-publisher -O http://pkg.opensolaris.org/dev opensolaris.org
pfexec pkg image-update

Will update if I run into any other issues or think of anything I’ve missed.

Update: 26/12/09:

The snapshot manager service has been placed offline due to a dependency problem

This message is displayed when trying open the Time Slider GUI interface. Look at the processes it lists, they will look something like:

maintenance svc:/system/filesystem/zfs/auto-snapshot:frequent
maintenance svc:/system/filesystem/zfs/auto-snapshot:daily

For each one run pfexec svcadm clear svc:/system/filesystem/zfs/auto-snapshot:TIMEPERIOD

So in my example, run:

pfexec svcadm clear svc:/system/filesystem/zfs/auto-snapshot:frequent
pfexec svcadm clear svc:/system/filesystem/zfs/auto-snapshot:daily

Start, restart, stop apache

Slipped my mind to mention this earlier, this is a bit unintuitive for Debian-people used to /etc/init.d/apache2

pfexec svcadm COMMAND http:apache22

Examples:

pfexec svcadm start http:apache22
pfexec svcadm restart http:apache22
pfexec svcadm stop http:apache22

Useless boot environment images

Another one that slipped my mind. pkg image-update creates new boot environments, assuming you don’t need the old images and want the diskspace back:

pfexec beadm destroy BENAME

Example:

pfexec beadm destroy opensolaris-1

View ZFS snapshots and their disk-usage

I tend to get paranoid as to how much space my ZFS auto-snapshots are using:

zfs list -t snapshot

Update: 11/01/10

Package ‘xrender’, required by ‘cairo’, not found

You get this if you when requesting details of GTK libs via pkg-config, eg. pkg-config --libs gtkmm-2.4. You need xorg-headers; pfexec pkg install SUNWxorg-headers. You will get this error trying to install Gtk2 for Perl.

Update: 07/02/10

Do you have a solution to an OpenSolaris problem you had? Feel free to leave a comment and I’ll edit it into this post.

I switched to Ubuntu around about April last year, migrating from Debian (and a little bit of Windows). Too many brain-dead decisions* by Canonical since then mean I am looking for a new OS.

I’ve used a number of different systems while testing Rakudo and while choice is a Good Thing^TM it is also making this decision a bit difficult.

The first thing to decide is if I will stick with Linux (change distros) or move to a completely different operating system. The Linux field looks like this:

Debian: Easy to use and is stable. I used to think apt was the one-true package management system but after having Ubuntu demonstrate how terrible it is as if you want a package update you have to rely on a third-party for it (just one example: not upgrading Firefox-3.0 to Firefox-3.5 — instead they had to be installed parallel, 3.5 having no branding and a modified user-agent) I am not going to an apt-based system. And there is still the reasons I left Debian for Ubuntu to begin with (long release cycle, software re-branding, SSH-key vulnerability fail, etcetera, etcetera).

Fedora: Never used it, not sure what to expect other than it uses yum instead of APT.

Gentoo: I think if I stick with Linux I will be going to a distribution that is vastly different from Debian/Ubuntu and Gentoo fits the bill quite nicely; everything is compiled. That said, I’ve done a Gentoo installation three times now; once on VirtualBox, once in a chroot to an external harddrive from Ubuntu and once on a spare machine. The problem is it takes too long; getting all my machines using it will be a chore and when it eventually breaks reinstalling will be another chore. Each time I’ve given up trawling through kernel config menus and used genkernel but that defeats the purpose but I am sure that if I used the config menu I will miss options that I need.

GoboLinux: Gobo is an interesting distro; it doesn’t use the normal directory structure but rather turns it on its head and uses paths like /Users/Username, /Applications/Firefox/3.5/ etc. Their package management philosophy is that the filesystem is the package manger; you install from source using a portage-like program called ‘Compile’ (they really should have given it a different name) and to uninstall you do rm -r /Applications/Foo/. In among their good idea are some bad ideas, one that really stands out is that the root user (which is called ‘gobo’) has their home directory on /Users/Gobo (rather than /root) :-/ And as interesting as it sounds I can’t help but think of Henry Spencer: “Those who don’t understand UNIX are condemned to reinvent it, poorly.”

Arch: Arch is a fun distro. It’s minimalist, pacman is nice, has a ports-like system (abs) which makes it like Gentoo without the painful installation procedure, and a rolling-release system (like Gentoo) which eliminates the apt-problem of being tied to third-party for updates. If I stick with Linux Arch will the be distro I use.

Outside of Linux, there is:

OpenSolaris: OpenSolaris has had a place on my harddrive for a while now. ZFS, TimeSlider, DTrace … it even looks amazing thanks to the Nimbus theme. However, it has its problems. The command differences are a bit of a barrier; pfexec (sudo), prstat (top), vmstat (free) but learning them won’t be too difficult and most of the tools are available with their original commands anyway (to increase/make adoption easier the GNU tools are shipped along with the Sun tools). It’s also slow, not detrimentally slower but still slower than Ubuntu. One of the major barriers is the driver support; neither sound or wireless work for me (although that isn’t a surprise). Wireless is a huge pain with alternative operating systems so I hope to get some rewiring done so I don’t have to use wireless but for now it makes OSOL unusable as a main operating system. That said, the truly great thing about OpenSolaris is Sun Studio. On OSOL, SS becomes a viable compiler set; on a platform where GCC has left its grubby marks, ie. Linux, SS binaries are slower. On an SS-compiled platform, GCC binaries are slower but SS binaries on an SS platform are faster than GCC binaries on a GCC platform. Which is a major win for OSOL to me. Definitely a contender of my new OS. (And no, Oracle are not going to kill it).

Haiku: Haiku is a fantastic BeOS clone. It’s a nice OS but I don’t think its for me full-time.

FreeBSD: The competition is between OpenSolaris and FreeBSD. Thanks to its licensing, FBSD is able to offer ZFS and DTrace (but still needs GCC). A major attraction is the Ports collection which is Gentoo’s system done right in my opinion (well, it was done right and then Gentoo copied it and got it wrong). Want a package fast? pkg_add -vr foobar. Want that package but compiled from source? cd /usr/ports/something/foobar && make. Great system.

So it’s between Arch Linux, OpenSolaris and FreeBSD.

Decisions, decisions.

Much the same as one upgraded from Vista to XP, I upgrade from Karmic to Jaunty today. The workarounds and hacks to fix everything that Just Worked^TM on Jaunty was too much — such as the DNS problem — everything I have tried to far resulted in working DNS but very slow internet, while in Jaunty it’s fine.

I need to come to a decision soon about what operating system to migrate to, problem is being stuck with wireless makes moving slightly more challenging.

(alternative title: Too Much Time on My Hands, which goes rather nicely with my last Perl 6 post).

So, in the beginning there was lolcats, then there was lolcode, then there was lolsql (jnthn even wrote a parser for it!) and then I took the joke too far and started lolsh, the lol Shell, in Perl 6.

It is very simple; it mimics bash’s UI as closely as possible, waits for input, and if the command entered is known it qqx’s the real-world equivalent.

carlin@cerberus:~$ cd Applications/rakudo
WTF?? I NO KNOE WUT U MEEN
carlin@cerberus:~$ CHANGEDUH Applications/rakudo
carlin@cerberus:~/Applications/rakudo$ perl Configure.pl
WTF?? I NO KNOE WUT U MEEN
carlin@cerberus:~/Applications/rakudo$ PURL Configure.pl
Reading configuration information from parrot_config …
Verifying Parrot installation…

carlin@cerberus:~/Applications/rakudo$ sudo make install
WTF?? I NO KNOE WUT U MEEN
carlin@cerberus:~/Applications/rakudo$ ICANHAZROOT MAEK install
[sudo] password for carlin:
/usr/bin/bin/parrot /usr/bin/lib/parrot/1.7.0-devel/library/PGE/Perl6Grammar.pbc \
…

Hopefully someone has a “so that’s how you do …” moment while looking at this so that writing it was justified

So a company that routinely removes functionality from their operating system, to avoid confusing their poor, helpless users with too many checkboxes and features, ships a release with broken DNS resolving?

I think, from an end-user point-of-view, “not breaking the internet” should be a lot higher up in the scale than helping out people that “don’t get files and folders”.

Or does Canonical hope that the same people that zapped themselves with ALT+CTRL+Backspace, failed to install important security updates (without an obnoxious prompt popping up every other day to remind them) and would be confused if the Firefox-3.5 package replaced the Firefox-3.0 package, will be comfortable and technically capable of manually editing their grub config to turn off IPv6. Assuming they can even get as far as to identify the problem. Or do they assume that the majority of Ubuntu users will be in the minority of people with routers that don’t splutter and die when they receive an IPv6 request?

Rant over. And in fact, I think this issue has helped me identify a problem that I was having with OpenSolaris, go figure

(Sorry for the bad pun).

One of the limitations of Mubot is that if karma is altered multiple times in a message, only the first alteration is actually made. For example, if someone says: John++ Cameron++ Cromartie-- Mubot only alters John’s karma (lambdabot would alter all three).

So after I had made the change to fix it, but before I restarted the bot, I demonstrated the current problem:

< carlin> mubot: karma mubot
< mubot> mubot is of an unknown quantity
< carlin> mubot++ mubot++
< carlin> mubot: karma mubot
< mubot> mubot has a karma of 1

Restarted the bot and proceeded to demonstrate the fix:

-!- mubot [n=zaslon@phoenix.theintersect.org] has joined #perl6
< carlin> mubot: karma mubot
< mubot> mubot has a karma of 1
< carlin> mubot++ mubot++
< carlin> mubot: karma mubot
-!- mubot [n=zaslon@phoenix.theintersect.org] has quit [Read error: 54
(Connection reset by peer)]

This is what the terminal said:

:sendak.freenode.net 366 mubot #perl6book :End of /NAMES list.
:carlin!n=carlin@phoenix.theintersect.org PRIVMSG #perl6 :mubot: karma mubot
:carlin!n=carlin@phoenix.theintersect.org PRIVMSG #perl6 :mubot++ mubot++
Segmentation fault

Worked fine after I restarted it:

< carlin> mubot++ mubot++
< carlin> mubot: karma mubot
< mubot> mubot has a karma of 3
< carlin> \o/

I believe the segfault was caused due to the size of the karma.log file (and subsequently the %.karma hash) as I imported every karma entry from lambdabot — About 1700 entries.

I might have to trim out some of the fluff, for example I don’t think we need to keep track of +++++[>+++++<-]>[>++<-]>'s 1 karma point …