Why did the Lemmy instances stopped upgrading the version they use?
Why did the Lemmy instances stopped upgrading the version they use?
It seems most instances still use version 0.19.3 and the only one using the up-to-date version is lemmy.ml. They used to update relatively fast. What's changed?
Given how long it took to fix the federation issues back in december (0.19 update), I'd be hesitant about updating, too.
Given how long it took to fix the federation issues back in december (0.19 update), I’d be hesitant about updating, too.
But you're on a 0.19.5 instance. Do you experience issues? There's some incompatibility with Mastodon, apparently, but given that 0.19.5 addresses serious privacy issues and has been out for a while without big problems, the privacy fixes are more important than keeping 100% Mastodon compatibility. Surely 0.19.6 will be out soon enough to address them.
At least I personally have not seen any issues in this current release.
0.19.4 requires upgrading Postgres (and exporting/importing the database to the new version).
I just don't have time for that right now since my Postgres DB is on a dedicated machine and is shared with a few other active services.
They make breaking changes like that in patch releases?? Do they not follow semver?
Lol, yeah.
I think they plan to do semvar, but if I recall (and I may be wrong), since it's pre 1.0.0, they're "allowed" to make breaking changes since it's still in alpha.
The lemmy devs are not exactly industry developers.
Hell, they don't even respect gpdr, and one day the EU is going to wreck an instance and maybe an admin too if they didn't separate liability properly.
- Anything can change while major version is zero
- Technically, semver only tracks public API changes
Technically this is entirely within semver since it's 0.x.y.
this isn't true. it was incorrectly stated in the upgrade guide but has been removed a while ago. it was supposed to be a recommendation due to some issues with postgres 15. there is no postgres upgrade required between 0.19 releases.
is shared with a few other active services.
Why would you do that, given that Lemmy is 0.x software?
Because I'm not made out of server resources lol. I've already got a beefy server that's dedicated to Postgres and is well tuned, and everything else already hooks into it. I've also had better performance (and less overhead) with one, big well-tuned database versus lots of stack-local databases.
If Lemmy goes tango uniform, then any damage would be limited to its schema. The worst it could really do would be resource starve it, but Zabbix would alert me of that quickly.
What are you looking at because the instance list shows lemmy.ml using a Beta version, and isn't the only one.
Dozens are using 19.5 which appears to be the latest stable version. Consider a couple of recent stable versions caused a lot of headaches for users and admins I think waiting to see if a newer 'stable' version is actually stable, let alone volunteering to test experimental versions in situ is prudent.
Just counted: 6 of the top 10 instances use 0.19.5 or newer. Source: lemmyverse.net with default sorting
We also upgraded fairly quickly and aside from some annoyances with the new image caching it's not worse than 0.19.3, so relatively speaking this was a good upgrade.
Lemmy.zip is on 0.19.5
Why did the Lemmy instances stopped upgrading the version they use?
They've done prod deployments in real life before and hence no longer believe in CD?
I wait a few days or so to see if they're planning on an emergency release to fix new bugs they didn't catch but I've found upgrading fairly easy, though I have a dedicated and fairly bog standard install and a smaller instance.
And Lemmy.ml is managed by one of the devs. And because this instance ist rather self entitled, they don't care about bugs that much
Wait we don't care about bugs because we're self entitled? Gonna need to elaborate on how that one works for me.