If you decide to run testing, be aware that while testing does get updates that address security issues after those updates work their way into testing from unstable, it does not explicitly get security updates.
Pinning some security sensitive packages to stable or unstable might be worth considering. E.g., if running testing on a client, pin firefox and extensions to stable
+ stable-security (note, globbing works too):
The above priorities will not downgrade to stable from testing if the packages are already installed. To downgrade priority needs to be >= 1000. See 'man 5 apt_preferences'. If priority >= 1000, probably best to only do that temporarily, then adjust to lower to prevent setting a landmine for your future-self.
If there are only a couple things you want to update to newer versions that are not in backports, you can just run stable, and pin those packages to versions in testing or unstable (but only if those packages pull in no / only a few dependencies not used by other packages). If you add e.g., testing/unstable sources to a stable system, add a catchall pin to force those packages to a low priority by default to prevent accidentally updating your entire system e.g., for sid:
Pinning without thinking can result in a broken system. But, I'm typing this on a box running testing (I guess stable, as of today) with packages pinned from Bullseye, Sid, and experimental and I've never had worse issues than an update being blocked due to dependency version conflict which was easily worked-around by pinning another package/removing or downgrading a pinned package; I run unattended-upgrades on all my boxes too. But, my hard rule is no scary deps e.g., a diff version of libc being pulled in, no deps shared with other packages that I would not want to have to pin to the same release (e.g., shared with any package with tons of deps itself), and no package that wants to pull in a lot of deps regardless of how benign they appear.
Pinning some security sensitive packages to stable or unstable might be worth considering. E.g., if running testing on a client, pin firefox and extensions to stable + stable-security (note, globbing works too):
The above priorities will not downgrade to stable from testing if the packages are already installed. To downgrade priority needs to be >= 1000. See 'man 5 apt_preferences'. If priority >= 1000, probably best to only do that temporarily, then adjust to lower to prevent setting a landmine for your future-self.If there are only a couple things you want to update to newer versions that are not in backports, you can just run stable, and pin those packages to versions in testing or unstable (but only if those packages pull in no / only a few dependencies not used by other packages). If you add e.g., testing/unstable sources to a stable system, add a catchall pin to force those packages to a low priority by default to prevent accidentally updating your entire system e.g., for sid:
Pinning without thinking can result in a broken system. But, I'm typing this on a box running testing (I guess stable, as of today) with packages pinned from Bullseye, Sid, and experimental and I've never had worse issues than an update being blocked due to dependency version conflict which was easily worked-around by pinning another package/removing or downgrading a pinned package; I run unattended-upgrades on all my boxes too. But, my hard rule is no scary deps e.g., a diff version of libc being pulled in, no deps shared with other packages that I would not want to have to pin to the same release (e.g., shared with any package with tons of deps itself), and no package that wants to pull in a lot of deps regardless of how benign they appear.