On Fri, 2022-01-07 at 11:00 +0100, Carlos E. R. wrote:
On 07/01/2022 10.47, Stefan Seyfried wrote:
So yes, paying customers really like old stuff :-)
Ok, that's about servers. What about desktop customers?
To some extent, they'll have to "bite the bullet" (I'm writing this on 6-months old laptop running Leap 15.3). This is what Leap is - an open derivate of an enterprise distibution. If you want the latest applications, flatpak, appimage or containers provide a wide range of possible approaches. I'm using flatpak a lot on Leap. (Yes, the integration of flatpak, appimage etc. could be smoother - ideally users wouldn't have to bother about configuring it at all).
There are people complaining that Leap doesn't run in their new laptops, doesn't this happen to paying customers?
There are various ways to make new hardware work with Leap. First of all, SUSE does a lot of driver backports for SLE, which will be available on Leap, too. Of course that's mostly for hardware that matters in the data center. Hardware vendors support SUSE with drivers and backports for server hardware, but not for home, desktop, or mobile devices. Thus storage and (wired) network devices should "just work" most of the time. For other devices, you can install the kernel from Kernel:stable or Kernel:stable:Backport, maintained by SUSE engineering, and enjoy every driver that upstream supports. And/or you can use add-on repos like "hardware" or "Printing" to get even more backports, KMPs, and user space drivers. (Maybe we should stop discouraging users from enabling these repos). Ideally, you get the best of worlds: A stable user space with support for recent hardware, timely delivery of security updates for the most crucial parts of the distro, and flatpak or similar for the cutting- edge stuff. That said, I would like us to push updates for user-space hardware "drivers" to SLE/Leap more aggressively. One example is the hplip printer/scanner driver suite, for which I am a maintainer. I see little or no reason to freeze hplip at old versions. The main reason for doing so is lack of QA resources, but hplip very rarely causes regressions, and even if does, they are hardly fatal. printd is another example, it's hopelessly outdated on Leap 15.3 but could be updated with minor risk (it will be in 15.4). There are lots of other examples.
I also appreciate stuff in Leap not changing as often as in Tumbleweed, in my desktop machine, laptop, and miniserver. TW is a no-no for me. But Leap is getting way too old.
Well, you've got to face the fact that Leap is like Ubuntu LTS, or even more enterprise-like. There simply is no product like non-LTS Ubuntu in the SUSE universe at this time. <rant> You should also take into account that a significant part of the problem is upstream. I used to be a python fanboy, but no more. I disliked how the py2->py3 transition was forced upon users, and dislike even more how backward compatibility is repeatedly busted in the 3.x series. If that didn't happen, this entire thread would be unnecessary. Yes, they announce these incompatibilities early. But still - they break stuff deliberately, and repeatedly. IMO python is not a good choice any more for mature projects with long maintenance cycles, unless the maintainers enjoy pointlessly "modernizing" their code every few months. </rant> Martin