If you skip every other Fedora version you have to (usually) upgrade once a year. And if you do the in-place upgrade (copy the VM before you do it so you can keep a copy of the old fedora around for troubleshooting) upgrades are pretty seamless as well.
I feel that about once a year, I would want to upgrade my OS version anyway, be it Fedora, Ubuntu or anything else (letâs face it: there is not much âanything elseâ in the ânormal person / non-technical expertâ desktop user market. Maybe Debian, but then you have the drawback of quite old software versions which people usually donât like. They want the shiny new stuff.)
An alternative could be to go with rolling releases but these can become quite unstable if users donât regularly install updates. I therefore consider them not suitable for the majority of users.
I see what youâre sayingâreleases are typically supported for 400 days or so, so if I install the latest one as soon as itâs released, I should be fine for a year. But releases donât immediately make it onto Qubes, so how long does it typically take for a new template to be released?
32 was released end-Apr. last year, templated mid-July last year, and has been EOL since the end of May.
33 was released end-Oct. last year, templated end-Feb., and will be EOL end-Nov.
34 was released end-Apr., templated mid-Nov., and will be EOL next May.
Someone who installed 32 as soon as it was released here had to install 33 or wind up using 32 past its EOL for six months if they wanted to skip it, which I donât think is good security practice (not a Fedora user; could be wrong). This is just one data point though, but basically what Iâm getting at is that once you factor in how long it takes a release to be made into a template, you canât skip like youâre suggesting.
Not a criticism of the time devs take to template a release, but I canât help but wonder if making two templates a year is worth the effort when Fedora use is essentially a historical accident like âQWERTYâ keyboards. Maybe converting something to templates doesnât require as much time and effort as I imagine.
I also feel that frequent Fedora EOLs is a trouble and time-waster. I solved it by using Fedora only for those qubes that do not require extensive customization. So I replace their templates every time and benefit from a refresh and compartmentalization for all those qubes every new version. For other qubes I am using Debian.
Just to be clear, I donât have a dog in this fightâI use a mix of Debian, microkernels, and other distros (hopefully OpenBSD soon). Issues with Fedora donât impact me materially.
I just see Fedora getting new releases and EOLs so often, I canât help but wince at the amount of migrating and rebuilding typical Fedora users must go throughâand I shudder to imagine what itâs like for those with many tiny templates for the extra compartmentalization. At least thereâs Salt.
Since Fedora is the default template, wouldnât this make Qubes even more challenging than it already is for newcomers? Wouldnât having Debian as default make more sense at least from this perspective?
I set all my Fedora VMs to the new F34 template. Almost everything seems to be working fine, only 1 issue. Sys-usb based on Fedora 33 could mount my external USB hard drive formatted in NTFS just file, without installing any special driver for NTFS. When using Fedora 34 it suddenly can not.
Regarding the Fedora vs Debian discussion, I actually prefer Fedora so I can use the latest software versions without having to wait for a new OS version.