On 08/09/15 14:41, Richard Brown wrote:
Hi gumb,
Thanks for your very long email, I dont have time right now to digest it all and address all the points I want to, but I want to snip right through and clear up a few points that stick out.
<snip>
Thanks for taking the time with this considered response.
The facts are pretty simple
#1 openSUSE's 32-bit user base is shrinking, fast #2 The current contributors for building the media want to support 64-bit only, because of Fact #1 #3 There are clearly passionate users who will be impacted by this decision (though they can continue to use openSUSE 13.1 and 13.2 for some years still)
If Evergreen sticks to its original plan (https://en.opensuse.org/openSUSE:Evergreen), 13.1 will be out of support around November next year. I hope it can continue until then - I know the Evergreen resources are small, but November next year is a lot more palatable for putting an upgrade solution in place for a couple of machines than November of this year. As for 13.2, that would normally be out of support come May 2016. So I'm not sure about those passionate users being able to use these for 'some years', unless they're willing to sacrifice network connectivity in the name of security. Another 14 months, perhaps, is all we've got. Actually I'm agreed that if anybody else can help out the Evergreen efforts, continuing to support 13.1 for at least 14 months more, perhaps even longer given the project's stated aim on that page linked above, that could be more efficient than putting efforts into a 32-bit release of Leap 42.1. I think part of what makes this change to 64-bit-only hard to digest is that it has come about as a result of the restructuring of openSUSE which was largely unforeseen at the time of the last 13.2 release. Users were informed that 13.2 was delayed due to the re-engineering within SUSE, as though it were a one-off exception (especially since one of the many inward-looking discussions on openSUSE releases only happened a couple of years ago, where it was decided to stick to the 8-month cycle and drop the .0 release). Many might have thought the following '13.3' release would follow eight or even six months later, in the circumstances. But this new tie-in to SUSE service pack releases has rather forced openSUSE's hand on the 64-bit issue. If things had remained under the original system, when the decision to drop 32-bit were made, it would likely have been announced in advance or even in conjunction with the previous release, a sort of 'note: this will be the last 32-bit version of openSUSE' statement. Instead - and this isn't a criticism of the new system which I think has some merit - we are rapidly approaching the next release and nobody had thought to make users aware that it would be 64-bit only. No conspiracy, no big fault of anybody, but unfortunately users are only now finding out shortly before a release, which has left people still on 32-bit-only little time to prepare.
<snip>
There's no reluctance to just spit it out. I admit to a little bit of a tiredness to repetition - I don't want to repeatedly go into great detail over a topic which has been discussed openly already.
For example, the statistics in question were part of my openSUSE Conference presentation
You can see the slides with the graphics here:
OK. I did watch your presentation at SUSECon months ago but the audio and slideshow were not easy to hear/see. Looking now at slide number 5, I was a bit confused at first until I realized that DVD+Net and Live are not supplementary downloads to the others denoted by architecture. So it looks like x86-64 was in the region of 105,000 downloads for 13.2, with x86 at about 29,000, or around 21.5%.
As a long-time openSUSE user, I've read a good number of articles, comments and threads in previous years from users in developing nations who say they obtain the distro on disc to install on multiple machines in their communities, and that for them high speed phone or internet connections are either unavailable or unaffordable. It would be logical to assume that people in these regions are likely to have a higher-than-average proportion of older and hence 32-bit machines. Are we no longer caring about this part of the userbase? openSUSE is embarking on its second Asia summit. Were it being held in one of the continent's less developed countries, imagine starting off with a keynote where all 32-bit users were asked to raise their hands, and then asked to leave the summit, because they're no longer relevant.
I was at the openSUSE Asia summit - all the hardware I saw used was 64-bit. I think it's actually quite narrow minded to think that developing countries must use ancient obsolete hardware - I think you'll find most of these countries embrace low-cost, but modern, hardware.
Well I know that in many European countries and the US there are organizations abound that cart off thousands of computers and devices to the developing world. They can't *all* be scams! Nonetheless, still a drop in the ocean I guess. I was quite careful however to distinguish in my remarks above that the country hosting the summit wasn't necessarily the same one that would have such older hardware. Neither China nor Taiwan represent the poorest countries in that region. And for those users enthused enough and with the technological prowess to attend such a summit, sure, they will not generally be the ones with the least resources or coming from the more remote or poor regions.
<snip>
Given that one of Linux's long-standing strong selling points has been that it runs very well on older hardware and allows you to keep such machines in use, for the sake of three more years of support for up to 15% of the user base, I think openSUSE should continue to offer 32-bit
But what if openSUSE has no contributors interested in working on it?
Well as far as 32-bit is concerned, it's clearly not a question of if. There are none, bar Per, perhaps.
And also, I continue to struggle with the mindset that somehow openSUSE must continue to find a way of run shiny new latest versions of everything on increasingly old hardware. New versions of openSUSE are an opportunity to provide our users with new technologies, new software, new versions.
There must comes a time that the desire to develop, improve, and move forward must mean that support for the past doesn't bring the benefit of the time, effort, and cost required.
openSUSE has also been about choice. Choice in desktop environments, choice in enterprise/community distro or now rolling release. Choice in configuring it just how you want it, and for whatever you want it for. But the focus of the distro has clearly now shifted. Somebody, in a thread I can no longer find, was just commenting on how openSUSE is now all about servers, cloud deployments, development tools and the like. That would fit in with the thrust of SUSE towards datacenters and software services. But it's a sharp contrast to where the project was at when I picked up my first boxed copy of 8.2 a decade ago. My eyes glazed over as I looked at the details and screenshots on the back of the box detailing the thousands of apps and useful tasks that could be achieved with openSUSE - very much oriented to the 'power desktop user' as well as developers. Even in the 11.x series the promotional material with the trendy guy in his big designer chair was aimed at the professional who wanted an easy life. The current target user seems to now be shifting, even if some people will still want to claim that openSUSE is everything for everybody. It's going more the way of the enterprise, the developer and the system administrator, and that level of choice and support for the desktop user and their stable of older machines is being eroded. You may care to disagree on that. Perhaps Tumbleweed is better suited to the average desktop user, so long as they have plenty of bandwidth, than an enterprisey LTS spin-off.
What would the business case be that I should present to SUSE?
SUSE are more than happy to invest in openSUSE, especially when it also brings benefits to themselves, but in the case of 32-bit support I cannot think of a pitch I could possibly bring to SUSE management that wouldn't be laughable
"Please give us money and/or more hardware so we can support an architecture which you already don't support in your own products, you're never going to support in your products, and we're going to have to absolutely have to stop supporting in a few years anyway at the current rate of decline? Oh and by the way, we currently have no contributors interested in building for this architecture"
But doesn't that exemplify a bigger problem, whereby openSUSE can only expect financial or hardware support if it aligns itself with the needs of SUSE and its enterprise product? If openSUSE is to be wholly independent, it should either get supported no matter what direction it wants to take, or it would have to just cut ties completely with SUSE as its sponsors and go seek funding elsewhere.
I don't want to see people leave, but if they really do only have 32-bit capable hardware right now, I'd imagine the cost and effort of upgrading to cheap, maybe even second hand, 64-bit hardware is no more difficult than moving to a different distribution
Well, in the case of my folks' machine, a reinstall of the root partition with Mint, preservation of /home and a bit of reconfiguration and learning of Mint's system tools, shouldn't cost anything except a few hours or a couple of days of my time. A new or secondhand machine will be at least 150 UK pounds (if we're talking about anything of remotely satisfactory quality on a par with their old but reliable existing machine). So if it really comes down to that, my choice is made. If I could keep them on a stable, updated machine and make that choice in a year from now, I might have had more chance to save up and help them with a new purchase.
Obviously, there are emotions to take into account, so such rational options will not be considered by all. Therefore, I'm sure we will loose some people over this. I don't like it, but I accept it's a cost that comes with progress and I really think it's more important we do what is right for the long term sustainability of the Project than burn ourselves out trying to make everyone happy in all cases.
I'm not against progress, but I can't help feeling openSUSE now has its hands tied on this issue as a consequence of aligning itself with the enterprise LTS base. And it may not be the last time that happens. gumb -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org