Web lists-archives.com

Re: Gitlab Evaluation & Migration




 ---- On Mon, 25 Feb 2019 09:54:26 -0700 Martin Flöser <mgraesslin@xxxxxxx> wrote ---- 
 > You asked for comments. I gave comment that I'm not pleased about yet  
 > another transition. Please keep that in mind. It means learning and  
 > interrupted workflows for every one. If you have already decided and  
 > don't want anybody to point out that transitions are painful, please  
 > don't ask for comments. Instead say that sysadmins decided. That's at  
 > least honest - your reply gives me the feeling the decision is already  
 > done and negative feedback was not expected. Sorry to feel this way. 

I think it goes without saying that transitions are painful, so it doesn't seem necessary to point this out because it's obvious to anyone who's ever undergone a transition. The point of any transition is that the pleasure of the new system should outweigh the temporary pain involved in transitioning, and the permanent the pain imposed by the current system.

Like you, I have some reservations about Gitlab. I'm not thrilled about losing approve/request changes statuses (that's in the EE edition only right now). And gitlab's fork-the-repo workflow feels more awkward to me than Phab's patch-based workflow, and I have some questions regarding scalability and the kdesrc-build experience given that each of our 80+ frameworks has its own repo. There are also some UI issues associated with inline comments in merge requests that I'd like to see fixed.

But there is a lot of pain currently associated with Phab. Multi-commit chains are very difficult to do correctly. Updating patches to reflect changes in the origin is error-prone. The website's user interface and search are poor. Landing patches is time-consuming and error-prone. The review experience for patches that change SVG icons is terrible, and occasionally suffers from a bug that that makes certain patches impossible to review [1].

All of these issues are fixed with Gitlab in my experience testing our our evaluation instance on https://invent.kde.org.  Have you had a chance to try it out yet? I didn't notice any comments from you in https://notes.kde.org/p/gitlab-evaluation-notes

And I think there's value in listening to the outside world when people tell us that they prefer Gitlab's  workflow to Phab's system that attempts (and fails) to abstract away Git itself. The enormous popularity of Github means that this workflow is at the very least  familiar to a huge amount of developers, if not outright superior. If we ignore that, we risk losing newcomers over time because our system is just weird and awkward compared to the norm, no matter how good our documentation is--and I think it's pretty darn good right now.

And yes, the documentation will go stale and need to be updated. I'll do this for Gitlab just like I did for Phab [2].


[1] https://secure.phabricator.com/T1022
[2] https://community.kde.org/index.php?title=Infrastructure/Phabricator&action=history).


Nate