Profile cover photo
Profile photo
Jamie Bennett
Technologist, programmer, researcher, tech evangelist, open source monkey, Linux developer and self confessed gadget freak.
Technologist, programmer, researcher, tech evangelist, open source monkey, Linux developer and self confessed gadget freak.

Jamie's interests
View all
Jamie's posts

Post has shared content
A great and productive time by all. Awesome to have the community around to sprint with Canonical on such an important topic and as a group, we really can make a difference. 
The Snappy Sprint in Heidelberg came to a close yesterday. Here are my thoughts, and a little background about the event, for those of you who've contacted me with questions.

First some background as to how this Sprint was organised. There were three tracks running, each broken down into sessions lasting between 45 mins to an hour. Each afternoon there was the option to participate in a 2 hour hacking session or continue to follow the other sessions. Trello was used to keep the session time table up to date. Wednesday evening we all had a meal together, lightning talks (mostly from the non-Ubuntu invitees) and an evening of hacking.

My first observation is the invited community contributors were warmly welcomed, had access to everything and were encouraged to participate, be it discussion, design or code. It was an inclusive and productive environment. ☺️

I mostly followed the community and cross distro track, but did dip in and out of some of the snapd, snappy core, +Snapcraft​ and snapweb sessions. From outside Canonical there was participation from AppSteam, +Arch Linux​, +Blue Systems​, Debian, +elementary​, +Fedora Project​, KDE, +MATE Desktop​, nextcloud, +openSUSE​, OpenWRT and VLC. And just to be clear, all the representatives are leaders and decision makers for their respective projects.

At the end of day one the discussions that had taken place in the community/cross distro track were described as "intense" by +Mark Shuttleworth​ in the daily wrap up. Linux distros have different missions and the desktop environments have different user experience stories. But even on day one, it was apparent we did all agree on something fundamental. Snaps enable us to deliver the best software we have to offer in a secure and predictable manner, everywhere.

On day two, the ideas started emerging as to how the different projects could leverage snaps and by the end of the week firm plans, initial code and in some cases final solutions were in place. The level of collaboration has been brilliant. The existing snap capabilities have been explored and some new limits discovered along with plans/proposals that outline how to overcome them. There was a particular focus on a new Snappy feature called Content Sharing, which is the mechanism by which shared platform/runtime snaps are made possible. A platform snap for GNOME 3.20 is nearly complete and the race is on to see if I can land the MATE 1.15 platform snap first 😉

I've left the Snappy Sprint with a clear understanding of the road ahead, an enthusiasm and commitment to be a part of the snap journey, and most importantly, some great new friendships. Keep your ear to the ground about snaps, there is certainly big news to follow in the not too distant future...


Post has shared content
Getting your desktop snap integrated in to the desktop environment just got a whole lot easier!

Post has attachment
Great snapcraft snapping tutorial by +Alan Pope 

Post has shared content
Listening to community feedback after the launch, we show you how to easily roll-out a vendor-independent snap store. Check it out on Dustin Kirkland's article!

Post has shared content
Awesome. Who said self contained packages have to be significantly bigger than their shared lib counterparts? Of course with snaps you get the added advantage of segmentation, security, and compartmentalisation.
The size of the LibreOffice snap is now 287MB, and here's why it wasn't to start with.

Post has shared content
Packaging hasn't been easy for app developers in the past but this has to be addressed if Linux platforms are to be a viable alternative to Windows and MacOS.
So lots of people are posting about the latest app package formats (AppImage, Snap, flatpack) and lots of other people then start flaming that this is all wrong and that there are all these issues with security and upgrades and size of the packages and what not.

So here is my rant:

If those app packages (for the record, I use AppImage for Subsurface) aren't the solution, then what is the solution?

The current situation with dozens of distributions, each with different rules, each with different versions of different libraries, some with certain libraries missing, each with different packaging tools and packaging formats... that basically tells app developers "go away, focus on platforms that care about applications".

Now I hear all the true believers in the true free software world howling in protest, telling me "you should love having to spend more time on packaging your app than on developing, you should be grateful that the brilliant packagers in the various distributions have packaged a two year old version of your app against a broken version of a key library and are still shipping it with all support requests going to you". And I admit, I should be grateful that all the people who think that running a computer is about deep knowledge and expertise and that the pain of most hardware not working is a rite of passage on that path to FREEDOM...

But in reality, most people are just looking for a computer that they can run a few apps on. Preferably the latest version of the app, the one that the developer supports, the one where the developer fixes bugs that they report.

If users cared about freedom they wouldn't be on Facebook (or G+), they wouldn't use iPhones, etc.

So the choice that I see is "make it possible for my users to get the app and run it on whatever OS their nephew installed on their system", or "give up on Linux and focus on making the app work well on Windows (right now about 70% of my users) and Mac (~20%)".

To be specific - I used to spend about three times as much time on issues around packaging Subsurface for Linux (~10% of the user base) than on Windows and Mac combined (~90% of the user base).
And for the past six months I have basically ignored the distros and just made AppImages available. I think right now my scripts still create Debian/Ubuntu package that may or may not work, the SUSE/Fedora builds are currently broken. And I really don't care - and it seems neither do the users. They just use the AppImage.

I'm all ears to hear about a solution that means that I spend about the same time for each of the platforms (for those challenged when it comes to math, that means at least 80% less work for all those Linux distros than it is today), and that makes things integrate well with the distros, allows for seamless updates, etc.

But if that exists, I haven't seen it.


Post has attachment
Are snaps really cross distro?

Post has attachment
Snap packages, which Canonical announced two months ago, will be the new standard for app distribution on Ubuntu and across whole host of other Linux distributions, including Arch, Fedora, and RHEL.

Post has shared content

Post has attachment
Contribute to a really exciting project - a step-by-step guide for potential snapd developers.
Wait while more posts are being loaded