Feel free to open an issue both for the http clone without .git, it seems interesting but I'm not sure how complicated it is.
Not sure about the git:// read only url, I don't oversee the consequences.
Re:Making GitLab faster
by jacobvosmaer on Friday February 26, 2016 @04:00AM
(#11180503)
The git:// protocol requires its own daemon (which is bundled with Git) running on port 9418 https://git-scm.com/book/en/v2/Git-on-the-Server-The-Protoco... . The standard Git daemon is unsuitable for GitLab I think because of the way it manages access control.
https://git-scm.com/book/en/v2/Git-on-the-Server-Git-Daemon
For this to work in GitLab we would need to find or write an alternative implementation of the protocol.
Sounds like a lot of work to me when we already have decent HTTP clone access.
Re:Making GitLab faster
by jacobvosmaer on Friday February 26, 2016 @04:04AM
(#11180511)
Hmm I was wrong, the current version of the built-in Git daemon has an --access-hook option so no problem there.
https://github.com/git/git/blob/56f37fda511e1615dc6df86c68f3...
The question is then: how badly do people want this and how many admins out there are willing to poke an extra hole in their firewall just for this protocol.
Re:Making GitLab faster
by jacobvosmaer on Friday February 26, 2016 @04:14AM
(#11180541)
Re:Making GitLab faster
by shadeslayer on Friday February 26, 2016 @05:33AM
(#11180731)
Here you go https://gitlab.com/gitlab-org/gitlab-ce/issues/13840
Thanks for the quick reply :)
Re:Making GitLab faster
by jobvandervoort on Friday February 26, 2016 @06:19AM
(#11180855)
Thanks! We'll follow up in the issue.
The only complaint I really have is the fact that I have to append a ".git" to the http url to clone the repository. Alternatively, it'd be nice to have a git:// read only url too.