I convinced my boss to set up @Forgejo and evaluate it. We are very unhappy with #GitLab and the current path we see. I already use forgejo for my private git. But a company has other needs, e.g. the actions and PRs.
I'm looking forward to see the outcome and hopefully switch to something better.
Florian Langer
in reply to Daniel Siepmann • • •Daniel Siepmann
in reply to Florian Langer • •My Boss (more into the hosting) has the info that there will be some change to docker in docker in the future, which will create some migration effort on our end.
My and my colleagues concerns are that we see GitLab becoming Atlassion 2.0, a bloated inconsistent beast of software. The UI/UX is already so inconsistent. The product seems to focus on so many things not important to us, loosing the focus on our important things. The UI is clunky for larger PRs. Also we don't get so many things while reviewing. It just doesn't make fun und is in our way, especially compared to UX on GitHub.
* Issues for example are that you can't easily mark a file as viewed and head over to next file within a PR. You need to click way too often with a mouse.
* Then viewed state doesn't seem always to be updated on pushes.
* Rebases won't properly work in MR views.
* Some components auto refresh, others don't, but I never know which one will and therefore constantly keep reloading different views.
* It has strange handling for some files in MRs.
* …
We only use the git hosting + MR (CI) part. No project management.
Does that help? :) And glad you are happy. There are so many ways and software solutions. Everyone should choose the one that works best.
Florian Langer likes this.
Daniel Siepmann
in reply to Daniel Siepmann • •@Florian Langer another personal thing: My issues on GitLab are more or less dead … nothing ever happened to them. I only had one issue with a breaking change regarding Forgejo. But they could help me within hours through their issues and improved the docs for the release and the handling within code within a day.
I personally prefer #OpenSource community projects over companies. You never know when their direction will change and how it might change. Focus might shift to AI instead of something valuable (I don't see much value in the AI, only a hype). Or they might sell to someone else that has completely other opinions and chooses another direction.
Unfrosted reshared this.
Florian Langer
in reply to Daniel Siepmann • • •Speaking of AI, it's indeed a lot of hype, but I do see a disruption upcoming for how people will search the web in the future.
That's also because most websites are just horrible to use or browse through (ads, consent pop ups, layout shifts on mobile etc.).
With the convenient way of just having an AI that does collect all this data for you and give you just the answers you need, I think this will be ultimately the new way.
Julian Hofmann
in reply to Florian Langer • • •* AI will stay ad-free/for free?
* content-publishers will grant free access for AIs?
@flanger @danielsiepmann
Florian Langer
in reply to Julian Hofmann • • •@Xitnelat
No, it will have paid tiers and / or ads. But I think it could still be very successful if done right. The dream of Apple, Google, Amazon etc. to provide an „Assistant“ that is actually helpful beyond setting timers.
And I think content publishers will try to fight against it of course, but in the end (if the people love the assistant) they need to comply eventually.
Btw I don’t say that I personally like this.
MarcusSchwemer
in reply to Julian Hofmann • • •@Xitnelat @flanger A word about quality ... it already declines
https://visualstudiomagazine.com/Articles/2024/01/25/copilot-research.aspx
New GitHub Copilot Research Finds 'Downward Pressure on Code Quality' -- Visual Studio Magazine
Visual Studio MagazineAlbrecht Köhnlein 🚀
in reply to Florian Langer • • •Daniel Siepmann likes this.
Mike
in reply to Daniel Siepmann • • •@flanger interesting reading others views and opinions.
I really like Gitlab (although I've used it for 10+ years, so have grown up with it). Agreed, there is bloat and inconsistent direction, but when I try to use GitHub I really struggle. Everything is tucked away and disguised in their minimalist design
Daniel Siepmann
in reply to Mike • •I've used bitbucket, bamboo, GitHub, Phabricator and GitLab, gitolite, gitea, forgeo and gerrit in the past.
I'm not too married to a specific software. And I'm always curious until I finally think that the software really fits. And I didn't don't that software for source code hosting + contribution or CI/CD yet. So still looking around.
I also don't think git is the best dvcs. There at other promising solutions out there.
But I know I can't change the industry...
Sydymar 🐧
in reply to Daniel Siepmann • • •GitHub - martinvonz/jj: A Git-compatible VCS that is both simple and powerful
GitHubDaniel Siepmann likes this.
Daniel Siepmann
in reply to Sydymar 🐧 • •Yeah I've learned about fossil recently and that's one of them. I like the idea to have the whole project offline and in sync.
I'll have a look at the other two. Thanks for sharing, highly appreciated.
Sydymar 🐧 likes this.
fnetX
in reply to Daniel Siepmann • • •Daniel Siepmann likes this.
Daniel Siepmann
in reply to fnetX • •Earl Warren
in reply to Daniel Siepmann • • •I've been asked by a small hosting company:
"Is Forgejo production ready?"
A simple yes/no question that is very difficult to answer. From their point of view I concluded that the most important aspect were:
* Is security taken seriously & communicated transparently?
* Is there a risk of data loss?
They are not very concerned about the UX as they only deal with the hosting part.
In February 2024 my answer was yes. Three months ago my answer would have been no.
Daniel Siepmann
in reply to Earl Warren • •Can you elaborate on the change of your answer?
I'm currently using forgejo for my own private projects. I never had any data loss and the security handling feels very good. Proper communication and releases.
So I'm curious why you changed your mind.
Earl Warren
in reply to Daniel Siepmann • • •Security has been everything one could expect from the start. The risk of regressions that could lead to data loss not so much.
Fixing that endemic problem is the primary benefit of Forgejo becoming a hard fork. This is what changed my mind.
https://forgejo.org/2024-02-forking-forward/
That does not mean Forgejo is 100% immune to regressions or dataloss. No software is. Only that the kind of regression that happened in summer 2023 can no longer happen.
https://forgejo.org/2023-08-release-v1-20-3-0/#fixing-the-risk-of-data-loss-related-to-storage-sections
Forgejo forks its own path forward
forgejo.orgDaniel Siepmann
in reply to Earl Warren • •Thanks :) I had an issue with that release because my combination of settings was not covered. But no data loss in my case. And I remember they had many info in place upfront.
Thanks for sharing :)
Earl Warren likes this.