[ehsankia]

Do people just randomly browse GitHub search for tutorials on how to use GitLab? That makes no sense. I'm pretty sure all the traffic that page is getting will be from news aggregators like HN, or from search engines, both of which don't care where it's hosted. It makes more sense to have it on GitLab so you're actually seeing the features being presented without having to go to a whole other site, and to show the site can handle the traffic / use-case.

[sov]

Well, then why does it even matter that it's hosted on GitHub? If GitHub is being used as little more than a public hosting page that people will directly link to, then what benefit is there from being on GitLab?

To be more specific about what I was saying earlier: the differences in form factor, features, layout, etc. are all going to be unfamiliar to those who don't know GitLab--so, presenting them a migration guide to GitLab from GitHub that's also already hosted on GitHub, with the layout and all the features they're already familiar with, is going to have a higher conversion factor than otherwise. It's not necessary that they randomly browse to this guide. Guides work because there's a smooth transition between the easy/familiar and the objective. By starting on the familiar, the guide is more inviting to follow.

[ehsankia]

I guess I can see it both ways, but in my mind, if GitLab's UI is too confusing for a new user to even navigate, then maybe it isn't good enough to migrate to. I do agree that familiarity with GitHub helps, but by that logic, a normal blog (maybe on Medium) would've been even better. The fact that it's formatted as a git repo (which is honestly sub-optimal for a tutorial), yet not hosted on GitLab, just felt strange.

[douglaswlance]

If the guide is on Gitlab, then you're in the Gitlab interface and immediately click around and open new tabs of the different things.

[amelius]

The README page should be obvious to any user, regardless of background. Sorry, but your argument doesn't make much sense.

[sov]

My argument isn't that the README won't be easily readable between git interfaces. My argument is mostly a) I expect that conversion rates will be higher going from GH -> GL if the guide is hosted on GH, and b) the guide is already hosted on GH, so, therefore c) you'll need a much better reason than a specious argument about dogfooding to compel a change.

I mean, it's just a git directory with a README.md--it's not like it can't be both places, but the argument from dogfooding I don't think is a good one.