RE: Lessons learned from the failure of Ubuntu Touch

Submitted by Scott Dowdle on Tue, 04/25/2017 - 17:37

I am NOT a fan of Bryan Lunduke.  It is a personal shortcoming of mine.  I watched him a few times when he was a host of the Linux Action Show.  That was a few year ago.  Anyway... I just found his know-it-all and smartass attitude annoying (perhaps I don't like the competetion?).  Then I attended one of his "Linux Sucks" presentations at LFNW a few years back.  Soooo annoying.  He sure has been milking that for years.  He is also a big social media person.  Ok, lots of stuff I loathe rolled into one.  I generally don't dislike people.  I mean, I don't even know the guy... but anyway.

Mr. Lunduke wrote an article for NetworkWorld (where he goes by Linux Tycoon) entitled, "Lessons learned from the failure of Ubuntu Touch".   Just look at his avatar picture.  Could he possibly be more pretentious?  In the article he makes the case that Canonical bit off more than it could chew by trying to make too many new things... which leads to his advice that the way to success for a Linux mobile project would be by starting small and then building on it.  Wow... that is "deep thinking".  Maybe he should change his name to Jack Handey.  But seriously, I believe he totally oversimplifies Canonical's goal and completely discounts their reason for having elaborate features.  The reason was to attract potential cellphone service providers who would want to license Ubuntu Touch and ship it on lots of devices... and then attract the hardware OEMs... or vice-versa.  Canonical's goal was definitely not for it to be self-installable by geeks.  While during the development phase they had to go that route... unless they had a flashy project they'd never be able to become "pre-installed".

Remember that kickstarter type campaign where Canonical wanted to design and build the smartphone itself... and do the software for it?  It ultimately failed although they did get quite a bit of pledges from true believers.  Just think what a nightmare that would have turned out to be if it had gotten backing?  I digress... and I mean no disrespect to Canonical or their fine employees.  It is just a big thing... and with that being their first hardware venture, a monumental task that was doomed to failure.  Canonical is better at software, right?

Being successful in the mobile space (for an OS, a system platform, or hardware vendor) is just as hard or perhaps harder than being successful in the gaming console market.  You pretty much have to have hundreds of millions or billions of dollars (pounds, whatever) to burn before you have a hope of succeeding.  You pretty much have to buy your way into the market.  And even with all of that, you might not be successful at gaining significant marketshare.  Just ask Microsoft.  How much did Google invest in Android before it started being successful?  It seems like forever ago but it was less than a decade, right?

I'm not saying a small, open source project can't be successful in mobile... but there are many different goal levels and meanings of success.  While Lunduke's advice on how to be successful might work for a certain group of open source developers, that was not what Canonical was going after.  They wanted to rival Apple and Android as much as possible... and show something that was impressive / to be proud of... not some beginner, we'll make it better later.. starter kit.  How do I know all of this?  I don't.  I'm just guessing... but at least I'm admitting it.

It would be nice to know where Mr. Shuttleworth thinks they went wrong... or what lessons they learned.  Now that would be interesting and revealing.  Hearing what someone completely outside the project (so far as I know) thought the problems were and how to fix them, not so interesting.

Video: LCA 2017 - Package Managers All the Way Down

Submitted by Scott Dowdle on Tue, 02/07/2017 - 14:28

Anyone who has been using Linux for a while is familiar with package managers and package management.  Being a Fedora user, I have noticed a few projects that Fedora has in the works to augment package management.  For example, Fedora Atomic does not the traditional package manager (dnf) but uses rpm-ostree instead.  Why would Fedora be working on additional packaging systems?  What is wrong with existing package managers?  I have been asking myself those questions for some time now.

Kristoffer Grönlund provides answers to those questions in a talk entitled, "Package Manages all the way down".  Enjoy.

FedBerry 25 on the Raspberry Pi 3

Submitted by Scott Dowdle on Sat, 01/21/2017 - 19:24

FebBerry LogoI've been running FedBerry 24 on my Rasperry Pi 3 for some time now.  It has been hooked up to the HDTV in the back bedroom.  While I don't use it on a daily basis, I do try to login to it once a week or so and keep it updated... and reboot whenever there is a kernel update. Given the rate of Fedora updates and frequent FedBerry kernel updates, the project is fairly active.

What is FedBerry?  FedBerry is a Fedora-remix made by three guys who have built glue packages for Fedora's ARM release and produced a few different images for the RPi 2 and Pi 3.  Download the .tar.xz, decompress and write it to a microSD card, insert card into Raspberry Pi... and snap... you have Fedora.  They started with Fedora 23, are on Fedora 24 now, and fairly recently released packages for Fedora 25... although no images for Fedora 25 yet.

As I write this, I'm in the middle of upgrading my F24 system to F25.  The number of packages the FedBerry folks have to produce is pretty small.  They are mostly related to the kernel and various branding packages.  It really isn't that far away from Fedora's ARM build.  Fedora has said that they are working on getting a release to run on the Raspberry Pi but historically there have been a few roadblocks that over time have been dissolving.  The main ones were with kernel support that wasn't in mainline and/or proprietary and the use of the FAT filesystem for the boot partition... or something like that.  I read a few blog posts on it a couple of months ago but don't remember the exact details.

Anyhoo, I run XFCE and a host of other common desktop software on the RPi3 and it works great.  While it is no speed demon, all of the hardware works including the wifi.  I can ssh into it and even connect to it via x2go for a remote XFCE session.  Overall, I'm very impressed with FedBerry.

I will update this post with info on how the upgrade went.  The FedBerry devs didn't announce their F25 packages, or at least not that I saw, but I noticed a 25 directory on their repo site and thought I'd stick my neck out.  If it fails on me, it would really be my fault for being a earlier-than-early adopter... but so far it seems to be working.

Update: The upgrade went fine.  Rebooted and had a 4.9.2 Linux kernel. FB24 had 4.4.41.  All the hardware continues to work fine.  Thanks FedBerry!

The method I used was: dnf system-upgrade download --nogpgcheck --releasever=25 followed by dnf system-upgrade reboot

Video: What's Next for Containers?

Submitted by Scott Dowdle on Wed, 11/02/2016 - 09:50

Red Hat's Vincent Batts gives a  presentation at systemd.conf 2016 conference entitled, "What's next for containers?".  It is a good overview of where the various container projects are (with no mention of OpenVZ however) and what work needs to be done.  I enjoyed his assessment that the first thing that is next is, "Get Past the Hype," and to, "Make Containers Boring."  Vincent goes over several of the userland tools as well as covers the areas where Linux native containers still need work.  Enjoy.

Video: systemd.conf 2016 - State of the Union / Portable Services

Submitted by Scott Dowdle on Wed, 11/02/2016 - 09:45

There have been a ton of conferences in the last couple of months... and luckily a lot of the presentations were recorded and have been posted.  Here is Lennart Poettering's presentation from the systemd.conf 2016 conference on, "State of the Union / Portable Services".  Enjoy.

Audio: Open Source Security Podcast, Episode 10

Submitted by Scott Dowdle on Tue, 10/25/2016 - 11:44

Everyone has been sweating the DirtyCOW bug that has been in the Linux kernel for over 9 years.  I recently discovered the Open Source Security Podcast and their most recent episode covers DirtyCOW some in a way I found interesting so I thought I'd share.  Enjoy!

Notice that is in Opus format which should play in Firefox and Google Chrome just fine... and supposedly Microsoft's Edge. Apple's Safari? Not so much