3 ms·

On the one hand, yes. On the other, why does Ubuntu always do their craziest changes in the LTS? The LTS should boring, and the one after it should be the wild

by larrik 2y ago

On the one hand, yes. On the other, why does Ubuntu always do their craziest changes in the LTS? The LTS should boring, and the one after it should be the wild one. (That may not be what the plan is here, but it does sound like it)

mort96 2y ago
I'm guessing it's because everything that ships in the LTS must be maintained for 12 years. I understand Canonical's desire to not still be maintainig Qt 5 in 2038. Tho the ideal time to make this change would be in 25.10.
- chris_wot 2y ago
  Perhaps they need to reconsider how an LTS works?
  - baq 2y ago
    In enterprise settings 12 years is not quite enough. There are racks of boxen set up with custom software targeting a particular distro version and there is no money or value in making changes to make it compatible with something newer. Hardware gets upgraded on 5-7 year cycles, but the software… isn’t happening.
    - throw16180339 2y ago
      As the saying goes, "Old hardware goes to the scrap heap, old software goes into production tonight."
    - hulitu 2y ago
      > Hardware gets upgraded on 5-7 year cycles, but the software… isn’t happening. Software gets rewritten every 1-3 years throwing away all the past experience. (Qt, GTK, and so on and so forth). /s
  - mort96 2y ago
    The value of LTS, and the reason companies spend tonnes of money on support contracts for those 12 years, is exactly this though: the set of libraries shipped in 26.04 will still be supported 12 years later without major backward incompatible changes. The very reason people would pay to keep using 26.04 in the mid 2030s and beyond would be that the software they built or bought in 2026 keeps working, typically without even having to recompile. You may argue that those customers should just update their software to use new versions of libraries or frameworks or replace discontinued libraries with new ones, or pester their vendor to do those things, or pay a bunch of money to consultancy companies to do those things; and you may well be right, but as long as those companies won't do that and are willing to instead pay Canonical, why wouldn't Canonical accept the money?
    - secabeen 2y ago
      Vendors don't always want to do this. I've seen 6-figure instruments stop working after a major update, and when we go to the vendor for an updated version of the software, their response is "we don't support that hardware anymore; buy a new instrument."
      - SoftTalker 2y ago
        And this is why RHEL4, NT4, OS/2, and other long-buried OSes are still in use in industrial applications.
- spockz 2y ago
  I think GP is intending to have it work as follows: If canonical would like to remove QT5 from LTS vN, they need to remove it in the first non-LTS release after LTS v(N-1). That way there is a lot of time before it to adopt which increases the chances of it actually happening at vN.

by larrik 2y ago

mort96 2y ago

I'm guessing it's because everything that ships in the LTS must be maintained for 12 years. I understand Canonical's desire to not still be maintainig Qt 5 in 2038. Tho the ideal time to make this change would be in 25.10.

chris_wot 2y ago
Perhaps they need to reconsider how an LTS works?
- baq 2y ago
  In enterprise settings 12 years is not quite enough. There are racks of boxen set up with custom software targeting a particular distro version and there is no money or value in making changes to make it compatible with something newer. Hardware gets upgraded on 5-7 year cycles, but the software… isn’t happening.
  - throw16180339 2y ago
    As the saying goes, "Old hardware goes to the scrap heap, old software goes into production tonight."
  - hulitu 2y ago
    > Hardware gets upgraded on 5-7 year cycles, but the software… isn’t happening. Software gets rewritten every 1-3 years throwing away all the past experience. (Qt, GTK, and so on and so forth). /s
- mort96 2y ago
  The value of LTS, and the reason companies spend tonnes of money on support contracts for those 12 years, is exactly this though: the set of libraries shipped in 26.04 will still be supported 12 years later without major backward incompatible changes. The very reason people would pay to keep using 26.04 in the mid 2030s and beyond would be that the software they built or bought in 2026 keeps working, typically without even having to recompile. You may argue that those customers should just update their software to use new versions of libraries or frameworks or replace discontinued libraries with new ones, or pester their vendor to do those things, or pay a bunch of money to consultancy companies to do those things; and you may well be right, but as long as those companies won't do that and are willing to instead pay Canonical, why wouldn't Canonical accept the money?
  - secabeen 2y ago
    Vendors don't always want to do this. I've seen 6-figure instruments stop working after a major update, and when we go to the vendor for an updated version of the software, their response is "we don't support that hardware anymore; buy a new instrument."
    - SoftTalker 2y ago
      And this is why RHEL4, NT4, OS/2, and other long-buried OSes are still in use in industrial applications.
spockz 2y ago
I think GP is intending to have it work as follows: If canonical would like to remove QT5 from LTS vN, they need to remove it in the first non-LTS release after LTS v(N-1). That way there is a lot of time before it to adopt which increases the chances of it actually happening at vN.