You're telling me there are cases where a company has so many projects in it's gitlab instance that employees routinely have to search for things? That seems odd to me. I'd expect there to be a naming scheme for most things, and even if there's some oddball project, I'd think somebody else you work with knows about it and where it is. My company has maybe 20-30 projects, and I can find any of them easily if I know the client's name.
Ever worked in a company with micro services or a company that has a bunch of employees? You'll have hundreds of projects easily if it's a bunch of teams. If someone tells me "it's in our SRE libs package" I'll go ahead and search that. (Usually I can't find it because it's in a separate namespace on our Gitlab instance. Doing that is an Enterprise only feature on Gitlab right now: https://docs.gitlab.com/ee/user/search/advanced_global_searc...)
The point is that search is important and not a social network gimmick.
Yep, same here. We architected our services so that a common layer could be used which reduces service duplication but still we have over 700 projects for just one company doing end-to-end ecommerce and fulfillment. At some point you even start running out of names so you end up with people using acronyms from their specific fields which quickly ends up being unfindable. Not because of search but because you no longer know what to search for.
Actually, it doesn't even have to be microservices. At my current workplace we have over 1000 projects in our Gitlab. And many of those projects are a collection of up to 20 libraries.
I work for a university that has a centralized IT, but each of the schools may also have an IT group. When they setup a local Gitlab instance I moved our team's (web development) projects over.
We have over 60 projects alone within our group (our own school was decentralized, and only in the last 5 or so years have I been able to get things centralized within our team), and there's a ton that the centralized IT group has on the instance, plus all of those that other schools share. We currently have three people in our team (two designers/front-end, one full stack), but at one point were at six plus a consultant.
I've got a good idea of what most of the 60+ repos contain, and where things are, but a good search for the various open repos, and those that I'm semi-associated with, would be fantastic.
Can someone explain to me what this does? It seems to me that it restricts Pages to project members, but I've had that feature enabled on private repos for at least a week or two now, which makes me think this is something different, but I can't figure out what.
Ah, okay, that makes sense, thanks! I can't tell you how much trouble I've had deploying docs with authentication to an org that uses GitHub, I basically spent a whole day figuring something out and gave up. GitLab was literally one setting.
It's nice to know other orgs also have these checklists where lots of items aren't checked, and you have no idea if they should be checked, or when, why, etc. Does anyone have a good solution to this?
I really like their checklist. It brought many details to my attention that I would have missed when developing such a feature. Does anyone know such a list of checklist (awesome-checklists or so)?
I love GitLab, and Pages access control is a great feature, but I think this was rolled out very poorly.
<rant>
The access level is "Only Project Members" by default for private projects, which I consider a breaking change because the steps I previously used to create public Pages (add .gitlab-ci.yml and push) no longer work.
The first time I tried to deploy Pages after the access control feature was enabled I wasted a lot of time because of this. When my new webpage redirect to the GitLab sign in page, I didn't bother signing in (why should I, when Pages have always been public?). I waited a day, because Pages have taken several hours to provision in the past[1]. Finally I started searching the web for why Pages was redirecting to the sign in page and found out about the access control feature.
I support having the access level "Only Project Members" by default, but I think the rollout could have been done much better. My main objections:
- There was no indication that the new webpage existed and that the issue was access control -- when I tried visiting the webpage for a non-existent project I got the same redirect. I understand why (to avoid leaking the names of private projects), but the redirected sign in page could have still shown a generic message ("project is missing or private") and mentioned that Pages are now private by default.
- The setting was not in an obvious location for me. I checked "Settings > Pages", which said the pages are served but did not give any indication that access control was enabled. There should have been a notice here saying that Pages are now private by default and that this can be changed in "Settings > General > Permissions", at least for the first few months after the rollout.
- The API [2] does not support changing the Pages access level yet, so I have to sign in to GitLab and change it manually (or fake the form submission). I want to be able to create a project with public Pages from the terminal, like I could before.
These issues could have stemmed from an assumption that developers heavily use the GitLab web interface and are always signed in. For me that is not the case.
Some small additional issues:
- The option "Pages access control" in Settings > General > Permissions is badly named because it's not clear what it means when it's toggled off. With the other options (e.g. "Issues", "Wiki", "Snippets") it's clear that toggling off the option removes the feature, but toggling off "Pages access control" could either mean "remove the access control feature" (making the pages available to everyone) or "remove the Pages feature" (making the pages available to no one). From my experiments it appears to be the second.
- The options have a glitch where toggling an option puts the corresponding access level in an inconsistent state. When the "Page access control" option of a private project is toggled off then on, the access level dropdown shows "Only Project Members" as selected but the value of the hidden <input> element is 30 (Everyone). Submitting the form sets the access level to "Everyone", as can be seen when the page refreshes. The other options have the same problem.
It's non of GitLabs business to play the judge in this. If a customer is doing something illegal and a court makes a judgment, they will probably do something.
I prefer a company not ruling their platform by their beliefs but by their countries laws and court rulings.
I think this is more nuanced. Right now companies cooperation with China is in the news. For example, both Blizzard and the NBA have betrayed American ideals in favor of making money in China. Other companies have built a dependent relationship with China over many years, and have been forced through this dependence to support China (Apple comes to mind here). Other companies are actively distancing themselves from politics (gitlab).
As a society, I think we need to decide what is best. In my opinion, companies should be entirely apolitical, both internationally and domestically. Further companies should not assert any influence _at all_ over political speech by their employees. In return, employees should not imply corporate support for a particular political view.
From my perspective, gitlab is doing the right thing. I don't think corporations have _any right_ to involve themselves in politics, even if the majority supports that position.
Edit: if popular sentiment is that American corporations shouldn't have a dependent relationship with China, then the _elected_ government should enforce sanctions. I don't need corporations enforcing moral policy.
Please keep in mind that "apoliticism" is in itself a political position: wanting to preserve the status quo and opposing change is still support for the status quo which, in itself, is a set of political beliefs.
It's impossible to be completely apolitical. Take your own example - blizzard - and tell me how they could stay moral without taking a political stance.
Morality is for people. Companies should just follow the laws of the country they are operating in. If we decide that isn't morally acceptable, then we can 1) stop using their services, and 2) push for change through our political representatives.
I agree that if Blizzard were to respond to the loss of customers by pulling out of the Chinese market, then that would be not entirely apolitical. Still, it is far less political than they are presently.
I don't think that article supports your statement.
Sure, that policy implies that Gitlab would take China's money if China wanted to be like any other customer. The policy doesn't seem to say whether or not Gitlab would take China's money if that meant censoring their other users.
