Downside to Automatic Updates

On the Debug 24.1 with Daniel Jalkut, Ryan Nielsen, and John Siracusa as guests, they talked quite a bit about what is going to happen with software updates in the future. I recommend both Debug 24 and 24.1 for a really great discussion around many things Apple and software development in general.

However, their almost wholesale support of automatic updates for everything scares me quite a bit. It isn’t surprising really because as a person who is excited about progress in technology myself I usually ride the edge as much as possible, but I think there are some things not considered by people who either are used to running the latest version of everything or have a vested interest in having people on the latest version of everything.

  • Automatic updates will allow developers and companies to push controversial features without people having the opportunity to keep an older version until everything cools off. Tweetie-turned-Twitter for iOS is a great example of this. If people were not able to keep their older version of Twitter for iOS, then you would have had to wait until it blew over or just grin and bear it because it is updated and you have no choice.
  • The idea of “tough luck” being the default answer for people who are automatically pushed to new versions seems like a user-hostile-but-developer-friendly way of handling things.
  • Being able to support anything goes out the window. If everything automatically updates then you no longer are able to keep anything stable. We already deal with this often with web applications and it is a pain (probably the greatest point of pain).

This is the part where people often scream “but this is the future, just deal with it” … and I smile, nod, and then think quietly to myself why.

As humans we seem to have a serious issue with thinking that something newer automatically is better. For whatever reason we think progress is always a good thing instead of always being a compromise each and every time and maybe, just maybe, it isn’t always a good thing.

We don’t always make things better. We don’t always get it right the second time around. It is entirely possible that the choices you are making today are worse in some very important ways and we need to live with that.

The every-present  feeling that we need to keep pushing forward faster and faster and into more areas is entirely human. However, we need to always be aware that while we might think we’re doing something awesome … we might just be wrong.


The Importance of HCLs

This is going to get a complete and total “DUH” from many of my system administrator friends out there, but I thought I would just share my experience as well … and why I should never just expect that a little extra work will be worth it in the end.

We run XenServer here on campus for our virtualization infrastructure and have been relatively pleased with how it has fit out needs. While updates should be trivial, a decision I made definitely complicated things far beyond what they needed to be.

For all other servers on campus we use Adaptec RAID controllers and have been very happy with them. We also used them in our storage boxes (running Ubuntu 12.04). They have been rock-solid and so I wanted to stick with what we know when it came to the XenServer host boxes as well.

The only problem is that Adaptec is not well-supported by XenServer. You can do it (and we did originally) by creating driver disks every time a new kernel is installed. That means that normal updates can cause headaches as you wait for the DDK to drop, then create the driver disk, and then hope that the driver installation works when you install the update. We did it a number of times and it worked fine, but it was not ideal.

It also caused an issue where we could not move from XenServer 6.1 to the new, open source, XenServer 6.2 at least in part because we were using Adaptec RAID controllers. That was enough. So I finally sat down with the HCLs for XenServer 6.0.2, XenServer 6.1, and XenServer 6.2 and worked to find a RAID card to help us with the upgrade to XenServer 6.2 and also push us into the future as well.

Needless to say, I found one, and I wish I would have done this sooner.

Working with a supported RAID controller has the following benefit, so far:

  • installation is faster because I do not need to install separate drivers for the RAID controllers
  • updates as faster because I do not need to have driver disks for each kernel update
  • upgrades are faster (and possible) because I can now do an upgrade to the XenServer pool without needing to do a fresh reinstall and migration of VMs
  • I have peace of mind because if something happens, I’m just a reinstall away from a working box again … regardless of whether I have the driver disks anymore or not

So I’ve learned my lesson. I am now going to pour over the HCL for a given software package before I go off making decisions. Fortunately we have use for the Adaptec cards I am replacing … and I am very happy to finally be rid of them soon in our XenServer host boxes.


My Problem with GNOME 3: Follow Up

Here is a follow-up to my earlier post. Thanks goes out to Phil Wels for getting me the needed screenshots so that I can finish the comparison.

These screenshots were gathered from an 11″ MacBook Air with OS X Mavericks and the latest Firefox. Since Apple added an explicit full-screen mode since OS X 10.7, I have two images for comparison (and you can find the GNOME 3 and GNOME 2 screenshots in the old post).



I can definitely understand why Apple added the full-screen mode. When you split things up you get the following numbers:

  • Windowed content height: 660 pixels
  • Full-screen content height: 704 pixels

That’s a big difference.

Even more significant (in my eyes) is that it validates my gut feeling from earlier. OS X, even when using a regular Firefox window, shows more content than GNOME 3 (and even a few pixels more than GNOME 2). When you take into consideration the full-screen Firefox, then it gets to be ridiculous.

If I did my math correctly, you get the following:

  • openSUSE 13.1 GNOME 3: 634 pixels or baseline
  • SLED 11 GNOME 2: 23 pixels or 3.63% more
  • OS X 10.9 (windowed): 26 pixels or 4.10% more
  • OS X 10.9 (full-screen): 70 pixels or 11.04% more

You can see the advantage full-screen has on OS X 10.9. Of course, this matters the most on the smallest of screens (in height), and 768 pixels is about as small as they come … but it does show much GNOME 3 does crowd out content more than GNOME 2 and OS X.


My Problem with GNOME 3

I’ve had openSUSE 13.1 installed on a Lenovo ThinkPad X220 for the past weeks and it has worked pretty well. I’ve used the GNOME desktop mainly because it has the ability to call up a search box with a single button (the Windows key on this laptop). That allows me really easy access to most functions I need to accomplish.

However, GNOME 3 has bugged me more than a little bit for one major reason and I’ll use some pictures to illustrate it.



Notice the subtle difference? Here are two more pictures to illustrate a little bit better.



It is really subtle and only noticeable on a small screen (like the X220’s 1366×768).

The usable area for a Firefox window in openSUSE 13.1 GNOME is 634 pixels high. On SUSE Linux Enterprise Desktop (SLED) it is 657 pixels high. While that difference of 23 pixels (give or take a few, I was just cropping the images in Acorn) might not seem like a lot, on a screen where pixels are at a premium it can make a sizable difference in just how it feels. The SLED desktop, at least in this instance, just feels a little less cramped.

Now, this is an extreme case, obviously, but it makes a difference. Smash such large toolbars on the top of the screen feels oppressive in a way that the SLED one does not. GNOME 3 overall just feels a little bit heavier than GNOME 2 at the moment on these screen sizes. On larger screens, obviously, the differences are not as drastic.

At this point I should state that GNOME has taken a decidedly Apple-ish approach here with Apple’s persistent toolbar along the top of the screens. I don’t have an exact analog with which to test it with to see how Firefox fares on, say, an 11″ MacBook Air which has the identical screen resolution. I’ll try to have a friend screenshot that soon so I can pull it apart.

The point of this was not to bag on GNOME 3, but to point out one difference I have noted so far between the openSUSE GNOME 3 and SLED GNOME 2 design directions and the feeling I have when using them. It is interesting, at least.

Business Technology

Agricultural Accounting

Well, Martens Family Farm is finally alive! After months of planning, meetings, more planning, more meetings and more meetings … the business is alive and we are pushing ahead with our plans for an expansion of the hog operation and also freeing up some more time for some partners so that we can continue to push things ahead into the future.

That’s the really exciting part.

The not-so-exciting part is that now we need to keep better track of where money is going, where it is coming from, and what it is being used for and why. That’s the nitty-gritty of what my job is from this point forward. I like to think of my job as “greasing the gears” to make sure that those who know how to farm can continue to do so without needing to worry as much about the money and compliance issues inherit with any business.

Alright, so I lied … that does sound really exciting to me.

What I need now is the ability to easily track income and expenses for an operation where I am only one of for partners and one of any number of people who are going to help on a small-but-growing family operation.

So I get to look at the current state of agricultural accounting software (or just small business accounting software in general, but for use within a farm).

Let me say, I’m not impressed so far.

QuickBooks is the “industry leader” at the moment and so far the online version as a few things going for it:

  1. Access from anywhere.
  2. iOS apps.

Alright, I lied, it really only has two things going for it. Other than that it is a highly complicated and slow web application which has 1000s of options to try to fit into any and all different industries.

I think I also need a basic accounting book to get my head around things.

However, I get the feeling that there has to be something better, easier, and simpler to use to get us started and off the ground. Having a mobile application so that I can easily add income and expenses right where I am would be a bonus.

On the cheap side of things, I have been thinking about trying to use GnuCash to get started. Seeing as it is free and doesn’t try to hem you into any set way of doing things, it looks like a decent lower-tech solution to our problem.

Needless to say I have a lot of work to do as we get things ramped up on our end. I’ll keep posting things as I find them, but I’ll leave with this plea to the software industry at-large:

There is a huge swath of farmers who are nearing retirement age and their younger progeny are going to be taking over the operations or at least taking a more active role in them. There is a huge opportunity to start developing software to serve this industry that doesn’t look like it was written for Windows 95. Do it. We are out there and we want to use software that is good.