The “bleeding edge” dilemma

The urge to run the latest trunk of KDE is too great ! but the fact that the trunk is not the most stable keeps a lot of us away from it. Our Friends over at project neon have done a great job trying to address the same issue.
But the problem is slightly bigger than this, imagine a scenario where you are working on two projects and for one piece of code to run you need to uninstall a few installed packages this breaks another piece of software that you need for another project.

Or another scenario would be , you are about to install an experimental piece of software. Do you leave your system in an inconsistent state if that experimental software doesn’t work ? Do you take a back-up ? or do you try to pull off a complex trick of lots of Alternate environments ?

The most embarrassing situation is when you go to your local Free Software User’s Group / LUG to give a talk about how great your OS and its applications are , and just when you are about to speak the app you were showcasing crashes , Damn !
How do you ensure that your apps don’t crash while giving the demo and also maintain your latest unstable hack to your favorite software without both getting into each others way ?

What is the solution to all of the above problems ?
I propose using Boot environments, on Open Indiana a.k.a Oi (Solaris) ( I have started to love this Distro , but my first love is always 🙂 )
Now supposing you have a clean install of you create a snapshot

$zfs create snapshot mydistro@firstInstall 

and now make this snapshot bootable

$beadm create -e mydistro@firstinstall FirstInstall 

Now go ahead and bork up your system 🙂
You may want to come back to the borked system again later, so you create a new snapshot and make it bootable.

$zfs create snapshot mydistro@borked
$beadm create -e mydistro@borked Borked 

now you want to boot into the borked Boot Environment next time.

$beadm activate Borked 

reboot, and in the OI grub you will find FirstInstall and Borked entries, select the one you want and continue working. FirstInstall lands you up
in the system before you borked it up. Enjoy hacking 🙂

more info about boot environments and beadm
That’s all for now! Give OI a try , you may switch over 😉
Until next time !

First step to getting KDE on OpenIndiana : Boot OpenIndiana ;)

OpenIndiana is a project created from the ashes of OpenSolaris. When oracle decided to close the development of OpenSolaris, some brave folks forked the kernel and put together a user-space to form OpenIndiana (OI), Its a great distro ! Above all its got software innovations (ZFS , dtrace, crossbow etc) which the linux userland can just dream of 🙂 (Though systemTap , btrfs are quite commendable as well.)

OK , so i have this great new distro and i love KDE so i want to see if my favorite softwares (Kate , amarok, digikam) work on OI or not. And sure they do ! (/me feels good here :)) the KDE – Solaris team has already created a repo with the latest KDE 4.6 packages. So , now i am convinced that i must start using OI. So i dump the dd image onto a USB stick and connect it to my laptop (a Dell Studio 1450). Now the fun stuff starts 🙂 OI doesn’t boot !!!!

After reading countless blogposts and consulting people on IRC (DrLou on #openindiana apparently had the same problem as me , and we sat together for 3 days (with breaks ofcourse 🙂 trying out everything that we could find) . In the mean time i had posted my problem to the OI discuss mailing list as well. Finally after a 2 day wait , i got replies from the OI gurus on the discuss mailing list. The message from Garrett D’Amore was particularly useful , he suggested me to add the -k option to the kernel line in grub and boot , this would fire up the In-Kernel Debugger the KMDB when i pressed “F1-A” during the boot process. But alas ! this didn’t work either 😦 Now i was desperately in need of some new ideas. Garrett’s post hinted me to check out more OI kernel flags. As it turned out, in an obscure wiki entry in BigAdmin (the solaris admin wiki) there was a mention of the flag “-B acpi-user-options=2” i appended the kernel line with this option and presto ! OI booted !. Now this was after 3 days of “beating around the bush” debugging and it felt great 🙂 . So, i headed over to #openindiana and told DrLou about my findings , and after a few tries even his laptop booted !! Yay ! (Supposedly this acpi bug comes to haunt users of solaris , from time to time :))

Now all that i had to do is to create a new beadm ( look up the solaris man pages , you will be amazed !) and add the bionicmutton repo to OI and that’s it, i have KDE 4.6 on OpenIndiana. OI Looks and Feels much better now 😉

This blog post is to document two things:
1. ” -B acpi-user-options=2″ might get OI to boot on your machine.
2. Never give up any problem , if you are facing it , you can be pretty sure someone else is facing it too ! stick to the problem and solve it.
(And after you solve it , in the Spirit of Good Will , share it with others by documenting it in your blog/wiki)

KDE solaris needs a lot of polish and could do with some more contributors. So if you have some spare time , drop by at #kde-solaris , #openindiana , #illumos on freenode.

That’s all for now ,
Until next time 😉

Urinal protocol vulnerability

When a guy goes into the bathroom, which urinal does he pick?  Most guys are familiar with the International Choice of Urinal Protocol.  It’s discussed at length elsewhere, but the basic premise is that the first guy picks an end urinal, and every subsequent guy chooses the urinal which puts him furthest from anyone else peeing.  At least one buffer urinal is required between any two guys or Awkwardness ensues.

Let’s take a look at the efficiency of this protocol at slotting everyone into acceptable urinals.  For some numbers of urinals, this protocol leads to efficient placement.  If there are five urinals, they fill up like this:

The first two guys take the end and the third guy takes the middle one.  At this point, the urinals are jammed — no further guys can pee without Awkwardness.  But it’s pretty efficient; over 50% of the urinals are used.

On the other hand, if there are seven urinals, they don’t fill up so efficiently:

There should be room for four guys to pee without Awkwardness, but because the third guy followed the protocol and chose the middle urinal, there are no options left for the fourth guy (he presumably pees in a stall or the sink).

For eight urinals, the protocol works better:

So a row of eight urinals has a better packing efficiency than a row of seven, and a row of five is better than either.

This leads us to a question: what is the general formula for the number of guys who will fill in N urinals if they all come in one at a time and follow the urinal protocol? One could write a simple recursive program to solve it, placing one guy at a time, but there’s also a closed-form expression.  If f(n) is the number of guys who can use n urinals, f(n) for n>2 is given by:

The protocol is vulnerable to producing inefficient results for some urinal counts.  Some numbers of urinals encourage efficient packing, and others encourage sparse packing.  If you graph the packing efficiency (f(n)/n), you get this:

This means that some large numbers of urinals will pack efficiently (50%) and some inefficiently (33%).  The ‘best’ number of urinals, corresponding to the peaks of the graph, are of the form:

The worst, on the other hand, are given by:

So, if you want people to pack efficiently into your urinals, there should be 3, 5, 9, 17, or 33 of them, and if you want to take advantage of the protocol to maximize awkwardness, there should be 4, 7, 13, or 25 of them.

These calculations suggest a few other hacks.  Guys: if you enter a bathroom with an awkward number of vacant urinals in a row, rather than taking one of the end ones, you can take one a third of the way down the line.  This will break the awkward row into two optimal rows, turning a worst-case scenario into a best-case one. On the other hand, say you want to create awkwardness.  If the bathroom has an unawkward number of urinals, you can pick one a third of the way in, transforming an optimal row into two awkward rows.

And, of course, if you want to make things really awkward, I suggest printing out this article and trying to explain it to the guy peeing next to you.

Discussion question: This is obviously a male-specific issue.  Can you think of any female-specific experiences that could benefit from some mathematical analysis, experiences which — being a dude — I might be unfamiliar with?  Alignments of periods with sequences of holidays? The patterns to those playground clapping rhymes? Whatever it is that goes on at slumber parties? Post your suggestions in the comments!

Edit: The protocol may not be international, but I’m calling it that anyway for acronym reasons.

ACK: This was originally published in XKCD