Apple threw the word 'magical' around a lot on the greatly punned and anticipated iPad. Is it really magical? I don't think even the most hardened fanboy would believe that. Instead, it's an incremental evolution of Apple's mobile platform. Bigger screen, more possibilities for things that the iPhone/iPod Touch are too daintily small for. Did much change around the operation system? Again, no, not really, just a minor evolution again with pop-over menus.

Will the iPad be a success? Sure, because Apple these days tends not to fail. Will it be a huge success? Likely not. I see the iPad fitting neatly with sales figures akin to what the MacBook Air is to the MacBook Pro. A specialized model for specialized tastes. By that yard stick, I think the people that genuinely have a need for a small, extremely portable media and light-work device will be extremely happy with it. Those that are perfectly happy with their laptop and iPod/iPhone combo won't see a need for it and likely won't "get" the device either.

It's easy to fault Apple for a lot of things but easiest to fault them for perceived "killer" features that the whole world sees is missing from newly announced devices and for whatever reason are left out by Apple. A few of these came to mind in watching the live feed of the event and I'd like to share my list with you.

#1 Wireless sync - The AppleTV has been out for quite a while now and the very first revision shipped with wireless syncing with iTunes. After all, it's not like you're going to disconnect the whole shebang from your HDTV just to get some movies on it. So they wrote and integrated very nicely wireless syncing for the AppleTV into iTunes. This has been notably missing from the iPod/iPhone primarily to be because they're not 802.11n devices and going over 802.11g would be painfully slow depending upon how much stuff you're transferring. Well, now we have a very mobile device that supports 802.11n and sync still has to take place over that damn white dock cable. There's no technical reason for it and it's bloody annoying. Support both syncing methods!

#2 Front-facing camera - Or any camera at all! The amount of real estate used by CCDs that go into mobile devices is extremely small and given how large the iPad is compared to the iPhone which does have a camera, there really is no excuse for not building in one. But this is the thing that has bugged me most about Apple mobile devices. Apple has done a fantastic job with iChat video and yet none of the mobile devices support it because they lack a front-facing camera which every MacBook on the planet has. To me, there is no bigger reason to have a truly mobile device than to be able to use it to communicate in whatever form. This could've been the iPad's killer-app. Completely free (with data access), face-to-face video conferencing on the go. It's a huge miss in my mind to ignore what they've had the ability and nuts-and-bolts to do for years now.

#3 Multitasking - One of the reason people adore their laptops is because they're mobile, full-fledged computers. I understand that Apple wants to focus on more scaled-down devices but to think that people really want to keep hitting the home button and stopping what they're doing to say check an e-mail is asinine. The iPad has the horsepower and battery life to do, so just let people be able to run multiple apps at a time, even if it's just 2 or 3. Think PIP perhaps?

#4 No flash - Flash is somewhat of a necessarily evil these days. It sucks CPU, it sucks battery but many websites use it in some form. You're really setting the iPad up to fail when you call it the best web browsing experience yet it still has little puzzle pieces where Flash embeds should be.

Those are my four biggest gripes, but really #2 is what hits home with me. I don't think I see a need to buy an iPad but, with my girlfriend and me both sitting on the couch with laptops open is evidence to, a little streamlining of web-browsing couldn't hurt either.

UPDATE: John Gruber has written a fascinating, excellent article on Apple and Flash for further reading on point #4.

I've said this for a very long time and I'll say it again: all cell phone providers suck! Period! There's no debate, there's no argument. Much of the fact can be attributed to that they simply service a vast number of people. Even the lesser known providers still service millions of people! Try serving millions of people one of three different types of cookies, you'll find that many people will be dissatisfied. To take the analogy further, try serving them those cookies while they're brittle and past-expiration while charging 4 times as much per month than the guy selling brownies down the street.

That jumbled mess of an example is a prime exhibit of why cell providers are pummeled to oblivion in the court of public opinion. It's not that they do that horrible of a job, it's just that the pool of people to please is so large that it's easy to anger hundreds, thousands and, most times, tens of thousands of people with one misstep. But the missteps are a plenty!

AT&T has dropped hints here and there of desires to follow their european brethren and start charging tiered rates for mobile data access. That is to say, offer plans for varying amounts of data usage, similar to how they grossly overcharge today for SMS usage. Yep! Think America is still the land of the supersized and excessed? It won't be for long. The proof is in the pudding as they say.

It actually started a year or two ago with the "Hey, me too!" goldrush of luggage fees with the airlines. Did you like the American concept of you-paid-for-a-ticket-your-bag-is-free? Too bad. We're going to take a lesson from the Europeans and every little thing is a nickel-and-dime-me-to-death extra! This is all done of course under the guise of lower base costs. And that is all bullshit as it's all smoke and mirrors. For years, American airlines watched as carriers in other countries charged fees for -any- baggage customers have. Camouflaged by outrageous gas prices in the summer of '08, the American airline industry made their move. One by one, they all instituted fees for the most basic of things including soft drinks and carrying a bag of stuff with you. The push from the airline industry to the communications industry is thinly veiled but is very much of the same vein.

Cable operators have wanted to do the same thing. They've been actively engaged in test pilots of tiered usage models and excessive usage cut-offs for years. All the while they advertise their service as unlimited. It seriously makes me physically ill. The whole point of this post however, is to tell you that unsurprisingly mobile operators are now making this very same push. The marching drum beats the same rhythm in all of the above cases. They tell you: "It's for your own good", "People will only pay for what they use", "It gives every one a higher quality of service". Again, it's all bullshit. The point of caps is to cut you off when you drink too much from the fountain because you'll cost them more than you may be giving them. All under the promise of 'unlimited', mind you. People under the employ of AT&T have dropped hints as to the intent of doing this at some point in the future.

"Operation Chokehold" as it is being called started as a lark. "Fake Steve Jobs" is a satirical column by Newsweeks' Dan Lyons. In last week's edition, he tells would-be readers that in order to teach AT&T a lesson for providing sub-par service, we must co-ordinatingly use as much bandwidth as we can at a certain point in time. This will proposedly show AT&T that their customers will not be intimidated into submission and that they vehemently oppose any such restrictions. AT&T and the FCC, for their part, is crying foul, noting that any such intentional attempts to disrupt communications are dangerous to the public good.

There certainly is some truth to their statement. Realistically though, right now there a few thousand Facebook "fans" of the operation. If a few thousand people (out of the many tens of millions AT&T subscribers) can completely or even in isolated locations take down the network, AT&T seriously has bigger problems. They need to think about their network design as a whole in that case and make in far more resilient than it is now. Think about it for a moment! If another disaster strikes the U.S., be it terrorist or not, what is the first thing 75% of ALL citizens are going to do? If you guessed "reach for their cell phone and communicate (through voice or data which AT&T says is the same thing essentially) with their loved ones", then congratulations, you're the big winner!

With everything in mind, who is right in this case? The people who just want what they are promised and are tired of greedy corporations that squeeze the last cent out of every customer that crosses their path or the corporation that has built their business model on specific constants of mob mentality that mustn't be broken for fear of structural collapse? When is it okay to say enough is enough and display some civil disobedience? How far can that disobedience go? What if it can potentially endanger a few lives for the trivial complacency of thousands, if not millions?

I leave that to you and whether or not you participate in "Operation Chokehold".

For years, I was totally happy to have a minimal, uninformative shell (bash) prompt. It wasn't that I didn't like information at my finger tips every time I'd press enter. I just found that most of the information got in the way of what I was trying to do. Then a few years ago, a co-worker put together (at least I assume he did, he might have found it elsewhere) a very informative and attractive shell prompt that disclosed the boxes vitals every time a command returned. Now, it's almost a necessity for me.

What makes a good shell prompt a good shell prompt? Well, first and foremost, it should disclose what box you actually are on and what is your present working directory is. Having these two things is probably the single most important way to avoid disaster since 90% of the time, you're logged into multiple boxes doing multiple things and knowing where you are typing on and in what context is essential! Additional data points are fluff, but having load averages can sure be handy when you're wondering why a box is nearly unresponsive.

Beauty is, of course, in the eye of the beholder. In my experience, ANSI colors make the terminal a better place to be and make it easier to identify different data types or points. Using colors in this way in your shell prompt help aesthetics and your ability to identify what information you're looking at. My favorite prompt uses colors liberally and sometimes takes people off guard when they rubberneck on my screen but I assure you it has as much function as form.

Obviously, the best prompt for you is decided by you alone but I'll share what my best prompt is. I've used this prompt on Linux for years and with my recent FreeBSD setup, I tweaked it to work on that as well. Here's what my prompt looks like...

For non-root:

For root:

What we have above is a good amount of information in each prompt that is of the second that the prompt is generated (upon completion of your last command). From left to right: user@host, load averages 1m/5m/15m, memory usage, uptime, ISO date, followed by pwd on the next line. The use of colors makes each value clearly identifiable and also easy to tell whether you are root or not. I've found this prompt highly useful in a number of scenarios, like say when you're looking back at your scrollback for exact times of incidents and recovery measures.

I put all the setup for this prompt in my /etc/bash/bashrc for Linux. I'll share this file in its entirety with you at the end of this post but really the prompt setup comprises of a few parts...

1) Define your colors. You need a palette to work with and this gives you the ANSI control codes to pull them off in your shell...

        BLACK="\[\033[0;30m\]"
        RED="\[\033[0;31m\]"
        GREEN="\[\033[0;32m\]"
        BROWN="\[\033[0;33m\]"
        BLUE="\[\033[0;34m\]"
        PURPLE="\[\033[0;35m\]"
        CYAN="\[\033[0;36m\]"
        LIGHT_GRAY="\[\033[0;37m\]"
        GRAY="\[\033[1;30m\]"
        LIGHT_RED="\[\033[1;31m\]"
        LIGHT_GREEN="\[\033[1;32m\]"
        YELLOW="\[\033[1;33m\]"
        LIGHT_BLUE="\[\033[1;34m\]"
        LIGHT_PURPLE="\[\033[1;35m\]"
        LIGHT_CYAN="\[\033[1;36m\]"
        WHITE="\[\033[1;37m\]"
        UNDERLINE="\[\033[4m\]"
        NO_COLOR="\[\033[0m\]"

2) Now that you have variables defined for the colors, you can then define your preference for colors for parts of the prompt and also the difference between a root and non-root prompt:

        C_L1=$RED
        C_L5=$GREEN
        C_L15=$BROWN
        C_PROC=$CYAN

        C_UPTIME=$PURPLE

        if [[ ${EUID} == 0 ]] ; then
                C_PRI=$LIGHT_RED
                C_PARENS=$BROWN
        else
                C_PRI=$LIGHT_GREEN
                C_PARENS=$LIGHT_BLUE
        fi

3) Finally, set the meat of it. Our PS1, the prompt itself:

PSHOST=`uname -n`

PS1="\n$C_PRI-$C_PARENS[$C_PRI\u$C_PARENS@$C_PRI$PSHOST$C_PARENS]$C_PRI-$C_PARENS[$C_L1\$(cut -d' ' -f1 /proc/loadavg)$C_PARENS/$C_L5\$(cut -d' ' -f2 /proc/loadavg)$C_PARENS/$C_L15\$(cut -d' ' -f3 /proc/loadavg)$C_PARENS]$C_PRI-$C_PARENS$C_PROC\$(free | xargs echo | awk '{print(int(((\$9+\$20-\$13)*100)/(\$19+\$8))\"%\")}')$C_PARENS$C_PRI-$C_PARENS$C_UPTIME\$(U=\$(cut -d'.' -f1 /proc/uptime);echo \$((\$U / 86400))d\$((\$U / 3600 % 24))h\$((\$U / 60 % 60))m)$C_PARENS$C_PRI-$C_PARENS$C_PRI\D{%Y-%m-%d}${C_PARENS}T${C_PRI}\\t$C_PARENS$C_PRI-\n$C_PRI-$C_PARENS[$C_PRI\w$C_PARENS:$C_PRI\\$""$C_PARENS]$C_PRI- $NO_COLOR"

4) Though this will not directly work on FreeBSD, you'll have to have linprocfs mounted and use this slightly tweaked version:

PS1="\n$C_PRI-$C_PARENS[$C_PRI\u$C_PARENS@$C_PRI$PSHOST$C_PARENS]$C_PRI-$C_PARENS[$C_L1\$(cut -d' ' -f1 /compat/linux/proc/loadavg)$C_PARENS/$C_L5\$(cut -d' ' -f2 /compat/linux/proc/loadavg)$C_PARENS/$C_L15\$(cut -d' ' -f3 /compat/linux/proc/loadavg)$C_PARENS]$C_PRI-$C_PARENS$C_PROC\$(cat /compat/linux/proc/meminfo | xargs echo | awk '{print(int(((\$9+\$20-\$13)*100)/(\$19+\$8))\"%\")}')$C_PARENS$C_PRI-$C_PARENS$C_UPTIME\$(U=\$(cut -d'.' -f1 /compat/linux/proc/uptime);echo \$((\$U / 86400))d\$((\$U / 3600 % 24))h\$((\$U / 60 % 60))m)$C_PARENS$C_PRI-$C_PARENS$C_PRI\D{%Y-%m-%d}${C_PARENS}T${C_PRI}\\t$C_PARENS$C_PRI-\n$C_PRI-$C_PARENS[$C_PRI\w$C_PARENS:$C_PRI\\$""$C_PARENS]$C_PRI- $NO_COLOR"

And that's it. I hope this has inspired you to use a better, more versatile prompt than what you typically get with your Linux/Unix distribution which typically just gives you a host and a pwd.

Linux /etc/bash/bashrc with cool prompt

I started out using Unix all of around 13 years ago or so. It's been a while. I had a shell account on Sun SparcStation 5 running SunOS 4 for a while. That's where I learned most of the basics while running an Eggdrop. It was basic, but it was enough to get the taste in my mouth. When it finally came time to get my hands dirtier, I chose FreeBSD to install on my own hardware at home. I wish I could say it was for some technical reason, but it really wasn't. Sure, they had the cute devil mascot. But mostly it was because, in 1996, Linux just wasn't there.

Linux in 1996 was still on the 1.x branch and transitioning to 2.0, arguably where the kernel got to a decent enough state in terms of stability and usability. FreeBSD was fairly friendly to install and manage compared to the Linux distros of the time. So that's the route I picked. After all, if it was good enough to run Yahoo, it was probably good enough for me. So I happily ran it for years on my home and co-located servers. It was rock-solid and supported all of the hardware of the era with ease. I also posit that the strength of FreeBSD is that it is fairly resistant to change over time, for better or worse.

Around 2004, I started to abandon FreeBSD for a much easier to manage Gentoo Linux. There were a few reasons for this, but the biggest one was portage. Similar to FreeBSD's own 'ports', it was far easier to maintain custom feature sets in software that was installed and WAY easier to keep everything up-to-date or at bare minimum, easier to visualize. Fast forward to almost 2010 and all my servers have been Gentoo for many years.

I haven't run a Linux server at home for a while due to power and heat concerns and left my serving up to my Mac Pro. But some mysterious crashing has led me back to the dedicated server model at home. I just so happened to have some spare hardware lying around, as I'm wont to do. So I decided to migrate my 2TB drives to it. The big question, of course, was what operating system to use. My choice centered heavily around what filesystems are supported since this would be primarily a file-server. Having used ZFS for years now, it seemed like a no-brainer. I would have given Btrfs a try if it were more mature and deemed at least 'stable', but it's currently still young and being actively developed. So if I wanted a FS that supported snapshots, copy-on-write, block-level checksums, ZFS is still the only game in town.

Naturally, when you say ZFS, the first OS you'd use would be Solaris or OpenSolaris, since Sun wrote the filesystem. So OpenSolaris it was. I popped in the disk and booted into the LiveCD. OpenSolaris gives you a hardware compatibility application right on the desktop of the LiveCD. It will scan your hardware and tell you what driver it's supported by. The problem comes when the driver isn't found. It does tell you when there is hope that your out-of-the-box unsupported hardware has a third-party driver, as it did with my Linksys Gigabit Ethernet card (RealTek 8169). The ever-so-handy Solaris NIC driver page had drivers for that. Unfortunately, the deal killer when it comes to Solaris/OpenSolaris on generic hardware is always hardware compatibility and driver support. This came in the form of the 3ware SATA controllers I use to get 8 SATA ports into one server.

There are no drivers for the older 3ware cards (specifically the 8006/8506 models) for Solaris. There are plenty of people optimistically asking if they're available on forums, but no drivers nonetheless. So what's a boy to do? Want ZFS, have odd hardware. The answer is obvious now in the last year or so. Use FreeBSD.

FreeBSD has supported ZFS since 7.0. Granted from what I'm told, it has gotten significantly more stable with 8.0. So that's what I did this holiday weekend. I loaded up FreeBSD 8.0 and much like when I originally purchased these cards for this purpose, they run flawlessly under FreeBSD. The RealTek card is of course supported by default as well. It's ironic in many ways for me. FreeBSD hasn't changed on the surface since my days of using 2.2.5 all those years ago. It still has the same (ghastly) menu interface for sysinstall. It still is color-blind by default. I'm sure it has changed under-the-hood by massive amounts over the years however. But the big plus now, is that besides Solaris, is the only other OS that supports ZFS with the default OS install.

It's still as rock-solid as I remember it. Even running ZFS, a memory hog by anyone's standards, it's still very frugal and runs easily on 1GB of RAM. Solaris would have a rough time doing the same thing. FreeBSD, while not being anywhere close to Linux in terms of hardware compatibility, still will run on gobs more hardware configurations than OpenSolaris. I'm more surprised than I'd like to admit, but I shouldn't be. Will FreeBSD become my preferred OS once-more? Well, there's a proper tool for every job and FreeBSD definitely doesn't fit every task but for solid-servers that don't need to be toyed with constantly, I think it's the winner.

I'm really saddened that in today's world, good software dies a terrible death because of the technicalities of the legal system. Licensing is a double-edged sword that anybody that writes a single line of code must heed. Not only licensing but its stalwart companion, patenting. I cringe whenever I hear either of those two words because innovation is ultimately the victim of both's murderous appetite for control.

What I'm really talking about today is the latest blow to ZFS, Sun's elegant solution to some of the file system world's oldest problems. Apple has officially ended the years of flirtation it has had with the file system. And with this step, in one quick motion, Apple might have doomed ZFS to an early death by relegating it to what appears to be a dying operating system maintained by a company soon-to-be absorbed by a not-so-sleeping giant.

Sun wanted ZFS to be open and adopted by any other OS that saw the benefit in it. They open-sourced it and generally promoted the hell out of it as a feature of Solaris. Many an OS have tried, but only one has succeeded at including it, that being FreeBSD. The porting it appears is the easy part. The licensing/legality aspect is where everything falls apart. The Linux kernel maintainers refuse to add it because the CDDL license ZFS is released under is incompatible with the GPL. That all seemed to put the spotlight on that "other Unix" out there.

Apple was enthusiastic about ZFS because it seemed to mesh up well with the goals of their Time Machine back-up system. There was legitimate serious talk pre-Leopard that ZFS could be the default filesystem. The evidence was there, there was a read-only ZFS implementation in beta releases of Mac OS X Leopard. And that for the most part is how things have stagnated over the past few years, with an open source project dedicated to fully porting the ZFS to OS X and apparently having complete support inside Apple.

What really was the nail in the coffin? We may not know for a very long time. Likely, it was somewhere between licensing and the current software patent lawsuit between Sun and NetApp. Either way, its certain that it was a technical challenge or issue. It was all down to the paper work, where great ideas go to die. I'm not sad because ZFS didn't win the crown, only because it lost on a total technicality.

This post serves more as a personal reminder to me than anything else. I ran into a situation at work where I couldn't figure out why sendmail on a Solaris 10 system wasn't using the 'smart host' that I could clearly see was in /etc/mail/sendmail.cf. Long story short, by default, sendmail on Solaris 10 uses /etc/mail/local.cf. You have to tell it you really want to use sendmail.cf. It serves a purpose though because a default install won't have the mailer daemon hanging out there for the rest of the world to probe (even if it won't be an open relay).

To make your intention known that you want to use sendmail.cf, you must edit the smtp:sendmail service properties. You can easily do that with svccfg. Like so:

svccfg -s smtp:sendmail setprop config/local_only=false

Then you have to refresh the instance to reload the xml files for the service...

svcadm refresh smtp:sendmail

And then all is well. The service should be restarted and using sendmail.cf!

Remember that announced date of "end of Summer" for MMS messaging on the iPhone? I do, to my displeasure. I have never been a fan of MMS. The reason is because I'm not a particular fan of paying for a duplicated feature. An MMS message is essentially an e-mail containing multimedia attachment(s) with an accompanying SMS notification. The visibility of these messages because they're presented somewhat like an SMS is nice but is it worth paying $0.50 or whatever insanity the carrier is charging for something that would otherwise be completely free via e-mail? My vote is no.

And then there was the announcement of iPhone OS 3.0 complete with basic features that had been missing on the iPhone since its inception (MMS and tethering, I'm looking at you). The presentation was rather forcefully adorned with logos of whom would be partners with providing these services. It was all but apparent it was to shame Apple's US partner: AT&T. And the shame should flow in AT&T's general direction. For they were not listed as being able to support either of these services on 3.0 launch day. In fact, they have yet to announce ANY iPhone tethering plans at all to this day.

MMS is not a groundbreaking new service. AT&T has offered it for years on practically all phones after year 2000. So why is it that hard to turn on a service for a specific device. It's likely because they explicitly had to turn this feature OFF for iPhones because they lacked support the SMS/MMS voodoo until 3.0. But really, how hard is this to do? Certainly it doesn't require 6 months to implement a simple profile change. The only thing I can come up with is that they expect a majority of iPhone users will use MMS messaging thus overloading their MMS (read: e-mail) servers. Again, 6 months is a hugely excessive amount of time for an infrastructure upgrade. What's more, we, the customer, get no feedback on this from AT&T aside from an estimate of "soon". AT&T is not Apple. They do not need to be a black-box, nor is it ingrained into their culture like Apple.

I suspect we'll just get a blitz from AT&T when the switch is finally flipped. That is a bit of a patronizing stance for a company to take but of any kind of company to expect this from, a cellular service provider is at the top of my list. Also, maybe they're just following Apple's lead (ala Snow Leopard), but at least Apple was early for once.

UPDATE: Lo and behold. AT&T has announced September 25th as the magic day. They also touch that infrastructure upgrades are in fact the reason for the delay. Go figure.

UPDATE 2: Now AT&T has even released a video intended to quell our fears or at least, offer up some half-baked excuses. 'Seth the blogger guy' = FAIL!

Apple released Snow Leopard on Friday. I've been using it for a few weeks now since it was announced to have gone gold. Does it live up to the hype? Wait a minute. What hype? This was the least hyped release of OS X I've ever experienced (I've been a user since 10.2, Jaguar). There is a good reason for that.

Mac OS X 10.6 is an incremental upgrade in many ways. It lacks the boat-load of new features that most Mac users have become accustomed to over the years. We knew for about two years now that this release would be a little anti-climatic. On the other hand, Apple has polished areas of the OS that were already good and turned them into shining beacons of what an OS should be. In short, $29 is a small price to pay for speed, stability and compatibility and that's what you get with Snow Leopard.

The problem is that many Mac users will left out of the Snow with this release being Intel-only. PowerPCs will soon be relegated to the domain of having an obsoleted OS if  you want to stick with OS X. This will enable still somewhat powerful PowerPC-based Macs to be had for pretty cheap giving you some cool hardware to run Linux on. Keep an eye on eBay!

Where does Mac OS X go from here? Will there be a 10.7 or are we finally drawing a line at release 10 point 6 and dreaming up major changes for the next MAJOR release? ZFS as a default filesystem, please. Transparent Windows binary compatibility, please. Some other crazy features that Apple is good at convincing us we always needed but never had, please! I suspect we may hear something about it at next WWDC next June or at least know what direction we're heading.

The time immediately succeeding Mac OS X releases is always interesting because we don't know where we're headed from here. You're usually kept busy by all the new features that are provided in the fresh release, but to be honest, we're not going to be kept busy with Snow Leopard. We're going to smile every now and then by the little quirks and niceties in it, but there isn't really anything that will keep the Apple faithful intrigued over the next few years until the next major release, which I'm hoping to actually be major and not yet another 'minor' version number increment.