The flatpak creator pointed out that we should check rather for symbols/features, because some kernels may be highly modified, as it is in EL. https://github.com/flatpak/flatpak/issues/2551#issuecomment-456002791
> Kernel requirements are tricky things. Generally the dependency is not on a specific version but on some feature/behaviour, and those are often backported to enterprise kernel without bumping the kernel version.
You can also perform some additional verification for runtimes and apps, to make sure that they won't require Linux > 3.2. Maybe CI that runs apps (and runtimes) on old kernels would help us to prevent problems like this in the future. This was originally proposed by Matthias Clasen from Fedora Silverblue.
Anyway, without additional checks, we can hit this problem again in the future (e.g. core20, core22, core24, etc).
> Is there something we can improve on the snapd side? /github. com/flatpak/ flatpak/ issues/ 2551#issue- 397366787
When we hit a similar problem last year in flatpak, I proposed to add an option to determine kernel requirements.
https:/
The flatpak creator pointed out that we should check rather for symbols/features, because some kernels may be highly modified, as it is in EL. /github. com/flatpak/ flatpak/ issues/ 2551#issuecomme nt-456002791
https:/
> Kernel requirements are tricky things. Generally the dependency is not on a specific version but on some feature/behaviour, and those are often backported to enterprise kernel without bumping the kernel version.
You can also perform some additional verification for runtimes and apps, to make sure that they won't require Linux > 3.2. Maybe CI that runs apps (and runtimes) on old kernels would help us to prevent problems like this in the future. This was originally proposed by Matthias Clasen from Fedora Silverblue.
Anyway, without additional checks, we can hit this problem again in the future (e.g. core20, core22, core24, etc).