back

by deepaksurti·9y ago·view on hn ↗
>> 2:41 – GitLab values boring solutions: our product should be exceptional

Exceptional products have exceptional UX. Gitlab IMHO has the worst UX of all git based products out there, I much rather take BitBucket over Gitlab. I tried using Gitlab, but no, I would much rather pay the 7$ to GH for my private repos.

I sincerely hope they make an exceptional product. And 'should' better be 'must'!

5 comments
This is pretty subjective obviously.

I think the GitLab interface is pretty good, especially compared to GitHubs. I can't speak for BitBucket since I've only used it once, but I do remember having a hard time finding my way around it.

UX is objectively measured. I don't know why people think it's subjective.
Could you explain what the objective quantification of UX is?
Clicks per task. Search time. User errors. For starts.

Here's a start: http://measuringu.com/essential-metrics/

Isn't his just moving the goal post? Now the subjectivity is choosing what metrics to measure.

Are 10 easy clicks worse than 2 hard clicks? Is a long, easy experience better than a short stressful one? And so on. Couldn't that vary by each person's preferences?

By trained or by untrained people? Domain specific experts, or newbies? .And why "clicks"? For many tasks, the way to get them done realy quick is with keys...

I had to do some IT support for a doctor's office many years ago, and (unrelated to what I was doing), had a chance to look at the system the front desk lady used to manage the patients, appointments etc. and it used a "clunky" old text based system (perhaps even DOS based?). TO me, it looked utterly bizarre and incomprehensible. But man, the front desk lady was lightning fast doing anything with it. Of course she trained on the system for years, and just entered keypressed rapidly that didn't make sense to me, but achieved the desired result... So, zero clicks, low search times, no (for me visible) user errors... so I guess they never should have switched, as this was the "objectively optimal" UI?

Personally, I don't think so. I.e. I think the UI could have been better, and easier to learn. But of course it's also important to see how usable it is once you do learn it... I was at that doctor's office again, not so long ago, this time as a patient... new lady (the old one retired, I assume), new system... flashy graphics... everything done with clicking, I don't even know if there still are shortcuts, the new lady certainly didn't seem to use any. And it did feel slower overall -- though this is of course a subjective evaluation only, with many years between the two observations. Plus it's somewhat unfair to compare somebody who trained a decade on a system with somebody who for all I know started a week ago... but I guess that's exactly my point here...

I listed starting points. There's an entire science for this. https://en.m.wikipedia.org/wiki/Human%E2%80%93computer_inter...

It sounds like you'd find it interesting.

If UX was fully objectively measured, wouldn't Gitlab, Github, and Bitbucket all look and feel the same?
If performance was objectively measured, wouldn't BMW, Mercedes, and Corvettes all perform the same?
Yes, performance can be objectively measured but different companies choose different metrics to optimize. The choice stems from subjective preferences. Hyundai optimizes for price and caters to the middle class, whereas Bugatti optimizes for speed and caters to whatever their market is. So subjectivity is inherent in the metrics we choose.
... and you have your answer to the prior question.
Derailing the intent of your parent comment and then taking the higher ground...good job bud
An objective measure of metrics does not necessarily imply there is a single optimal solution.
You shouldn't make design decisions from case studies alone, but the subjective experience of each individual user is a perfectly valid measure of UX performance.
No it's not. That is completely against the idea of ux. What someone says is not what they do. Measure. Analyze. Experiment. Tweak. Results. It's all objective data.
GitLab may not be as good as some others in the space, but I've never seen a repo UI that actively repulses me nearly as much as BitBucket's. Even SourceForge doesn't put me off like that.

Honestly, all Atlassian products have terrible UI. JIRA is just as bad.

> Honestly, all Atlassian products have terrible UI. JIRA is just as bad.

The individual UIs are not that bad. The problem is that people tend to use not only e.g. JIRA, but also Confluence, or more. And then the (massive) differences between their products in UX appear.

As a sysadmin, there's one thing that really annoys me: all that Atlassian stuff is written in Java (which means: on every problem one has to dig through pages after pages of Java stacktraces), and it eats resources like nothing else. GitLab got better in the recent versions, but it also had its fair share of interesting problems with memory leaks.

Oh, and both don't support MySQL (Atlassian does, but very limited) - which is bad if nearly everything else is MySQL and you already have a MySQL DBA.

> The individual UIs are not that bad.

I've still gotta disagree. Jira's UI is horribly inconsistent and haphazard. Some actions are buttons, some are links, some are hidden until you mouse over, everything feels like it was just tacked on wherever there was (or really wasn't) free space. It's probably the worst UI in a modern, commercial web application that I've encountered.

It's pretty bad, but the list of contenders in "worst UI" is quite large.

It feels like they took some Photoshop files from a designer and gave it to their dev team and said "make this, I don't want to hear any complaints."

but it's agile, so it has to be good.
I personally like the Bitbucket UI. Except the useless overview page. The rest of the UI is great IMHO.
I'm curious what you dislike about the UX. Personally I like the "feel" of the website generally (though it's not as polished as GitHub) and just find it to be slow, which I think they're working on improving.
You are right, GitLab doesn't have a polished UX. Improving that is one of our main objectives for the next two months: https://about.gitlab.com/handbook/ux/strategy/

If you are curious about what we're doing, take a look at the issues under UI Polish: https://gitlab.com/gitlab-org/gitlab-ce/issues?scope=all&utf...

Or feel free to point out any particular concerns here or through the issue tracker: https://gitlab.com/gitlab-org/gitlab-ce/issues

Personally, not a fan of how a readme-less repository just has an entirely useless main page. I much prefer having the file view a'la github on that page. That's the main UX complaint for me. Still use gitlab bunches.

Main complaint overall is the incredibly sluggish git push speed.

You can change that on your personal prefs to show the file list by default, with Readme on bottom: http://c.ekin.io/3z1V2v3P041b

git push speed is fine on our own ce installation but gitlab.com is beyond shitty.

We're considering changing the default so people don't have to find the setting https://gitlab.com/gitlab-org/gitlab-ce/issues/26834
I'd +1 this as default behavior. I want to see the source code!
That is kind of a bizarre thing to be tucked away in settings =| Noted, though
Where else would you put it?
I think it should be the default setting. Even make it not configurable. Why would anyone want to see just Readme or just Files view when it's possible to have them both?
This is something that has been on our minds for a while, and we must do it better. We created an issue for a smart project dashboard that aims to provide a better solution, it would be great to have your input there: https://gitlab.com/gitlab-org/gitlab-ce/issues/27112
Speed is one of our main priorities for GitLab this year. We've started Gitaly, a project dedicated to that: https://gitlab.com/gitlab-org/gitaly

We expect it to become a standard component of GitLab in Q1 2017 and to reach full scope in Q3 2017.

Can't wait : )
Personally not a fan of the sidebar navigation. I keep getting it confused with the individual repository navigation bar on top.
Navigation is another thing we are focusing on for the next several months (https://about.gitlab.com/handbook/ux/strategy/). In the next release, we are simplifying it (https://gitlab.com/gitlab-org/gitlab-ce/issues/26200). We plan to continue to improve upon our navigation and information architecture. If you are curious, you can follow along some conversations that are happening here: https://gitlab.com/gitlab-org/gitlab-ce/issues/27557.
I've only spent a few minutes there, having signed up just a moment ago. What jumps out at me about the UX is that source code does not seem to be front and center like it is in Github and Bitbucket. The README is all you see on the first page of a project and I had to hunt a few seconds before I noticed the "files" link.
Thanks, I added your comment to https://gitlab.com/gitlab-org/gitlab-ce/issues/26834 where we're considering changing the default.
For me the main issue is that reading READMEs or any other markdown files is tiring and annoying because of the ~1200px width.
Gitlab only have mobile site. I want desktop version.
Hey Deepak, thanks for the feedback. We've built out a UX team and improving that aspect of the product is high on our list of priorities, but please feel free to reach out if you have any specific issues. Happy to document those for you.
Hey Amanda, I would list the following: 1. Intelligent profile: shows my top 6 active repositories, ability to mark work I do as salient so that shows up on the dashboard in summary instead of my recent issues/merge requests etc, the salient work marking should be automated at a future point of time (ML). I really think of this as an automated Developer Story in the long run. 2. Recent area: Where you show the contributions in the last year, recent issues/merge requests etc. 3. Code Reviews: A list view on the left which shows all the files changed and a cart where I can add multiple files (ex: a.py, b.py, c.py) to show how the changes my be inconsistent etc to a peer. A 4 quadrant view will be very handy.

These are the top 3 things I can think of, based on my interaction so far. I wish Gitlab success. Thanks.

> Gitlab IMHO has the worst UX of all git based products out there

Remember that git-based products include Gerrit (a UX which I hear hate about, but frankly, I like a lot)