Well, there are still quite a few bugs and regressions in NC28, and there were definitly more issues than with other major releases, or let’s say more serious ones involving basic features which affect many users. But I’m still confident that most of them will be resolved when 27.1 will be End of Life. Nextcloud 28.0.3 Milestone · GitHub
well this is just one opinion.
I understand why you might have that perception, particularly with v28.
But: v28 is a special beast. The entire front end was swapped out. A huge amount of technical debt addressed. It was a long time coming.
(And v26 and v27 are still actively supported; the latest major - particularly early in its cycle - is always for the early adopters).
While the pre-v28 UI appeared reliable and stable on the outside, it was a massive PITA to maintain, enhance, or fix bugs in.
There was no way around it (well other than maintaining the status quo but then all sorts of other enhancement work, bug fixes, and performance improvements would be stalled or consume more dev time than otherwise necessary).
Unfortunately, the front end changes in v28 were inevitably going to introduce issues due the newness of the code and visibility (since it’s the main UI).
The devs involved in the Files front end have been working their butts off to fix things as they are noticed. We’re only on v28.0.2. v28.0.3 - with >70 fixes and counting - is just around the corner (this month).
I’m just a hobby user running a small NC instance for 10 family members. I’m not a programmer so I can’t contribute much of anything there. Take it for what it’s worth but I can’t agree more with Dennis1993.
I originally came to NC looking for a self hosted alternative to google drive for file and picture sharing. Over the years we’ve expanded our use to include passwords app and the calendar app for shared calendars. We were really happy with NC!
But then things started to go down hill. First photos app, which worked very well for our us got “updated” which made all but useless, at least for us. There’s numerous complaints going back years on that. Nothing has really been done to address the shortcomings so we had no choice but to abandon using it.
NC added the dashboard for the web UI which was great. We used the calendars dashboard widget extensively to share group calendars and subscription calendars within the group. We really came to rely on it
That is until it stopped working. Same deal, there’s bug posts going back well over 2 years. The programmers keep working to add new “features” but there’s no movement to solve the old problems. We had no choice but to abandon the calendar dashboard widget too. That did not go over well with my users aka family. At least we could still go into the calendar itself and see the shared events though.
This last upgrade brought even more calendar problems as any external subscription now is no longer visible! Now the calendar app is all but useless for us.
Unfortunately with the photos dashboard and calendars no longer working we’re to the point of abandoning NC. I really hate it NC used to be a great solution but it seems there’s little interest in fixing the problems which a piling up faster than they’re being solved.
I’ve long since given up installing the major upgrade every few months, instead I run the version until it’s EOL and only install the security updates. Unless there’s something new that I can’t resist, of course.
I also always wait to see what the community says before I install the latest version number because sometimes there are situations that cause headaches and steal my time because there are updates afterwards anyway.
I have been working with Nextcloud since the project started and the software has become the most important tool for me but experience shows that with every version upgrade some apps stop working.
This is probably the biggest problem but nobody seems to care.
I would also like to see a move away from PHP and a complete rewrite of the software in Rust.
This would simultaneously eliminate minor but annoying bugs that have existed for a long time and would be a relief for everyone, because then you only have to install a single executable file on the server to make a basic installation of a Nextcloud instance.
Well, yes, this suggestion had to come. The “Rewrite in Rust” that automagically solves all the issues. I’m not a developer, but I’d still like to point out a few things:
OwnCloud did a complete rewrite in Go with Infinite Scale. They have been developing this for years and are now at version 4.0 and it has nowhere near fetaure parity with the regular OwnCloud, let alone with Nextcloud. Actually, it can only do file sharing and that’s it.
A complete rewrite is a huge task, especially for a project as big as Nextcloud, and it would probably take years to implement anywhere near the same functionality as it has now. Sure, Nextcloud probably has many more active contributors than OwnCloud, but they are not all Rust experts.
And yes Rust is known to have fewer bugs and security vulnerabilities than many other programming languages, and it helps developers to write more secure and reliable code. However, it doesn’t automatically write bug-free software by itself, so again, you first have to find enough people who know the language well enough and want to work on the project.
And at the same time, of course, you have to continue to maintain the existing software, which also ties up resources.
Nope. Rewriting NC in Rust means that NC would loose all apps and ship with its own webserver instead of NGINX/Apache. That is a lot of extra work for them just to make NC run and cover all the things webservers do right now. And they would loose all contributing developers. The new software would have other bugs that will never be solved since that is caused by lack of manpower.
Since Rust needs to be compiled (unlike PHP which is interpreted) NC & App developers would specifically have to support all the OS & Architecture combinations. Especially for Apps this would probably just end up with shipping the source files and then compiling them on the target machine. And then you can go figure out why it didn’t compile.
Thanks @Bill for the examples, that really helps to understand the problems.
Removing features is really bad, and then it is hard if you have to upgrade for the limited support cycle. I remember with ownCloud (before the fork), there were different contacts and calendar applications. It makes kind of sense to have a single one, however the others were created because of missing features.
Not sure about the right way, not sure if you can keep two systems alive until the new one is enough feature complete to make the switch?
Bugs with regressions should be handled with priority (I thought this was the case). And the regression bugs must be fixed before you have to upgrade to a newer version and you loose this functionality.
Sorry, but where are the examples? It’s all relatively vague and there are no links to any actual issues. The only thing he describes in at least some detail is the issue with the calendar subscriptions. However, if they are not visible at all, then this is certainly not a general bug, but a specific issue with his instance, or maybe with a specific version of the calendar app.
Why? What exactely doesn’t work anymore that worked before?
Sorry, but what does “no longer working” even mean? These things do generally work on my instance, although I don’t really use the Dashboard and the Photos app that much
In my mind, most of the updates are seamless so I have no complaint if from a version to version basis my experience is largely the same. The difficulty comes when managing PHP, you can run into a cascading effect where it is a major ordeal to update Nextcloud and if you are not careful you can ruin your installation.
For example, you need to update Nextcloud, which drops support for your version of PHP, then you go to update PHP and find that it is not in the Debian repositories and you have to update Debian!
Assuming you have a supported PHP in your repos, you now have to manually edit several config files which have hard coded PHP versions in them.
You also may make the mistake of jumping two versions at once which can break your Nextcloud, or in the case of jumping from Debian 10 to Debian 12 will ruin your installation and you will have to reinstall the whole OS.
Thus Nextcloud upgrades often carry a sense of dread that perhaps you will miss a detail and have a major ordeal to overcome.
It’s easy to handle PHP: get the snap or docker.
I installed the snap in 2018 and I haven’t touched anything ever since.
Sometimes I’m still amazed by how smooth things run so far.
The snap lack basic features like thumbnails for photos because of the intricacies of bundling ffmpeg and there are no real solutions for these problems. Much of the ethos of self hosting involves allowing customization which is not really possible with the snap. As for the Docker, one must learn and understand containers in general, which is something on my list to do eventually but it is not as simple as turning a key.
And all containerization does is kick the can upstream one level to someone else managing the issue.
maybe you wanna try AiO?
@soprus why not staying with snap if you are happy.
so… as this could get kind of offtopic I’d recommend staying back on topic.
@ bb77 - were are the examples?
" specific issue with his instance" that would flush if there wasn’t so many others with the same issue.
And hence the main problem with NC in general. The first response is always the “it’s you, not us” mentality. I’m done. No need to keep rolling around in the mud.
Regarding die photos app: Two questions. How many percent of the photo gallery apps in the app store are not able to sort pictures by date taken (from the beginning)? How long did it take the “new” photo app to get this feature?