Who’s responsible for bug reports on old software versions?
5 points by riki
5 points by riki
In theory, with Linux distributions are supposed to act as the first line of support with their bug trackers, doing triage. In practice, this rarely works out, because users either don't know about the downstream bug tracker, or don't trust it to resolve issues. Sometimes it's lack of resources on the downstream side as the article suggests, other times it's because distro policies mean stability may mean locking in existing bugs (cough cough, Debian). I know I personally don't really bother with distro downstream bug trackers (except for Gentoo) based on experience.
The exception is if you have a commercial relationship with your downstream distribution. I know at work, our clients do go through us first for our commercial PHP distribution.
I mean this is just an unsolvable problem.
There are upstreams who go "this is an old version, the bug is fixed, thanks" and then there's "this is an old version but the bug is unfixed, thanks" and then there's this train wreck and everything in the spectrum between.
Users will give bad bug reports, or shout "$software is broken" on social media without even noticing they're using a 3y old version. And some software is just moving so fast it's really bad to package.
Reading this I'm still 95% on the Mint maintainer's side though, jftr.
But the question "what do we do if the maintainers object to their software being packaged" is one I've not seen a good solution often.
This surely depends on the nature of the bug report? Security bugs are the most obvious ones to back deploy, but if a bug is fixed/not present in later releases how far back is there a reason to back deploy the fix?
If the fix is “update your software”, and the response is “no”, then what is a project meant to do? It logical result is that every bug fix in a later release must be back ported to all prior releases.
Security or data loss bugs would be the only real cases I would consider to warrant back deployment more than a couple releases back.
In practice this is a strong argument for using something like AppImage to bypass the distro altogether.