5 ms·

What does GitHub have that makes it irreplaceable in a short time? It has many many users that's for sure but unlike proprietary instant messengers or social n

What does GitHub have that makes it irreplaceable in a short time? It has many many users that's for sure but unlike proprietary instant messengers or social networks I don't see strong network effects.

PieterH 11y ago
There are three aspects I can see. One is firms that use github internally, who have gone through the pain of learning to manage it. As long as the license cost remains less than the cost of moving away, they'll stay. Two is organizations who have complex team setups. You'll have noticed GitHub's increasingly complex team management? This is sticky. The more you invest in clicking on the right buttons the more you feel committed to the platform. The third aspect is integration with external systems like Travis. We've come to depend on this as part of our process. Apart from those three cases, I think it's still relatively easy to switch to something like gitlab.
dudul 11y ago
I agree. Especially for OSS developers. Maybe some integrations are missing with tools to monitor repositories, or automatically download dependencies. I would actually be quite happy to see GH's monopoly fade a little. I personally don't like this company, but mostly I don't like the idea of one repo manager to rule them all.
dominotw 11y ago
>What does GitHub have that makes it irreplaceable in a short time? All the alternatives are slow, clunky and ugly.
- straws 11y ago
  And unopinionated. It takes a tremendous amount of listening, polishing, and gardening to arrive at what github has built.
mbrock 11y ago
I think the social effects of GitHub are very strong. It's the de facto arena for open source collaboration. By using it, a project signals that they choose the most widely used community, and thus encourage participation. It's familiar. Everybody has an account. There's built in GitHub support in package managers, even, so the namespace of "githubuser/project" is valuable. Personally I'm not interested in another clone of GitHub.com. Let's try another model entirely. Something more distributed and federated. That way lies diversity and immunity to lock-in. Let's collaborate in an open fashion using open protocols and decentralized hosting.
- dudul 11y ago
  Agreed. I also see very little value in simply cloning GitHubg, that's one of my problems with GitLab. I like it, it's open source and the perfs get better, but at the end of the day, it is a clone of GitHub. Not much innovation. I would love to say a player take some risk and try to present a new model and new way to leverage Git.
  - mwcampbell 11y ago
    How about a Git repository management application designed specifically for Sandstorm (https://sandstorm.io/ https://sandstorm.io/)? Maybe call it Gitgrains.
  - sytse 11y ago
    What kind of innovation would you like to see? We already added protected branches, integrated CI and many other features. Our current plans are on https://about.gitlab.com/direction/ https://about.gitlab.com/direction/
    - freeadvice 11y ago
      > What kind of innovation would you like to see "I would love to see a player take some risk and try to present a new model and new way to leverage Git." I may be wrong but doesn't gitolite predate GitLab and didn't that have branch protection? Plus Perforce, ClearCase and other enterprise SCM had this concept for decades. Don't take this response in the wrong way, as I would love for GitLab to succeed, but GitLab is definitely not leading or creating features that Atlassian and GitHub are eager to copy. Innovating is doing things that others have not thought off and for the innovation to be marketable, it has to be useful. I truly want GitLab to be successful, but it's currently stuck in the "Samsung is a ripoff of Apple" stage. In order for GitLab to break this stigma, they really need to focus on something that makes GitHub and Atlassian really need to copy them. From the directions page, I really don't see anything that I would consider to be innovative. The only two points from the directions page that I can somewhat say GitLab could be innovative, would be the analytics and that's because analytics is so abstract/vague. If GitLab could leverage it's continuous build information and reliably tell management how well an employee is performing and/or how buggy their software is going to be, then that would be innovative. If GitLab could develop some formula based on the information that they are able to capture, and accurately predict how things will be, would definitely be something GitHub and Atlassian would want to copy. Along with IBM, Microsoft, etc. Adding continuous integration isn't really that big of a deal considering how well Travis and others integrate with GitHub. Right now, GitHub is the pretty girl, with glaring short comings, but guys put up with her because she is really pretty. Atlassian is the businessman that can get things done and doesn't charge that much. GitLab is the geeky kid that nobody is really taking seriously. There is no denying that geeky kid is useful, but people really only go with it because it melds with their open source ideology and/or because it's the cheapest solution. GitLab needs to find a way to break this current perception. They can do one of three things: - Become the new pretty "it girl" - Become a serious business man - Be both If you can be both, you would definitely be the innovative one. But right now, GitLab is that gawky geeky kid and that perception needs to change. And the worst mistake you can make is believing you are the pretty girl, because other geeks are saying you are pretty. Or believing you are that serious business man, because other cheap business man are saying you are the best. Like I said, I really want GitLab to succeed, and in order for it to succeed, it really needs to be the one that comes up with the innovations. Asking others for what to innovate, is not innovating.
davexunit 11y ago
People are very locked into their pull request and issue tracking tools, unfortunately.
- merpnderp 11y ago
  I migrated a few projects to Gitlab and pull requests and issues seemed to come over with only one problem, all issues were created by me. Not a horrible experience, but could be better (and maybe if the original creators had gitlab accounts, they would have been correctly assigned).

PieterH 11y ago

There are three aspects I can see. One is firms that use github internally, who have gone through the pain of learning to manage it. As long as the license cost remains less than the cost of moving away, they'll stay. Two is organizations who have complex team setups. You'll have noticed GitHub's increasingly complex team management? This is sticky. The more you invest in clicking on the right buttons the more you feel committed to the platform. The third aspect is integration with external systems like Travis. We've come to depend on this as part of our process. Apart from those three cases, I think it's still relatively easy to switch to something like gitlab.

dudul 11y ago

I agree. Especially for OSS developers. Maybe some integrations are missing with tools to monitor repositories, or automatically download dependencies. I would actually be quite happy to see GH's monopoly fade a little. I personally don't like this company, but mostly I don't like the idea of one repo manager to rule them all.

dominotw 11y ago

>What does GitHub have that makes it irreplaceable in a short time? All the alternatives are slow, clunky and ugly.

straws 11y ago
And unopinionated. It takes a tremendous amount of listening, polishing, and gardening to arrive at what github has built.

mbrock 11y ago

I think the social effects of GitHub are very strong. It's the de facto arena for open source collaboration. By using it, a project signals that they choose the most widely used community, and thus encourage participation. It's familiar. Everybody has an account. There's built in GitHub support in package managers, even, so the namespace of "githubuser/project" is valuable. Personally I'm not interested in another clone of GitHub.com. Let's try another model entirely. Something more distributed and federated. That way lies diversity and immunity to lock-in. Let's collaborate in an open fashion using open protocols and decentralized hosting.

dudul 11y ago
Agreed. I also see very little value in simply cloning GitHubg, that's one of my problems with GitLab. I like it, it's open source and the perfs get better, but at the end of the day, it is a clone of GitHub. Not much innovation. I would love to say a player take some risk and try to present a new model and new way to leverage Git.
- mwcampbell 11y ago
  How about a Git repository management application designed specifically for Sandstorm (https://sandstorm.io/ https://sandstorm.io/)? Maybe call it Gitgrains.
- sytse 11y ago
  What kind of innovation would you like to see? We already added protected branches, integrated CI and many other features. Our current plans are on https://about.gitlab.com/direction/ https://about.gitlab.com/direction/
  - freeadvice 11y ago
    > What kind of innovation would you like to see "I would love to see a player take some risk and try to present a new model and new way to leverage Git." I may be wrong but doesn't gitolite predate GitLab and didn't that have branch protection? Plus Perforce, ClearCase and other enterprise SCM had this concept for decades. Don't take this response in the wrong way, as I would love for GitLab to succeed, but GitLab is definitely not leading or creating features that Atlassian and GitHub are eager to copy. Innovating is doing things that others have not thought off and for the innovation to be marketable, it has to be useful. I truly want GitLab to be successful, but it's currently stuck in the "Samsung is a ripoff of Apple" stage. In order for GitLab to break this stigma, they really need to focus on something that makes GitHub and Atlassian really need to copy them. From the directions page, I really don't see anything that I would consider to be innovative. The only two points from the directions page that I can somewhat say GitLab could be innovative, would be the analytics and that's because analytics is so abstract/vague. If GitLab could leverage it's continuous build information and reliably tell management how well an employee is performing and/or how buggy their software is going to be, then that would be innovative. If GitLab could develop some formula based on the information that they are able to capture, and accurately predict how things will be, would definitely be something GitHub and Atlassian would want to copy. Along with IBM, Microsoft, etc. Adding continuous integration isn't really that big of a deal considering how well Travis and others integrate with GitHub. Right now, GitHub is the pretty girl, with glaring short comings, but guys put up with her because she is really pretty. Atlassian is the businessman that can get things done and doesn't charge that much. GitLab is the geeky kid that nobody is really taking seriously. There is no denying that geeky kid is useful, but people really only go with it because it melds with their open source ideology and/or because it's the cheapest solution. GitLab needs to find a way to break this current perception. They can do one of three things: - Become the new pretty "it girl" - Become a serious business man - Be both If you can be both, you would definitely be the innovative one. But right now, GitLab is that gawky geeky kid and that perception needs to change. And the worst mistake you can make is believing you are the pretty girl, because other geeks are saying you are pretty. Or believing you are that serious business man, because other cheap business man are saying you are the best. Like I said, I really want GitLab to succeed, and in order for it to succeed, it really needs to be the one that comes up with the innovations. Asking others for what to innovate, is not innovating.

davexunit 11y ago

People are very locked into their pull request and issue tracking tools, unfortunately.

merpnderp 11y ago
I migrated a few projects to Gitlab and pull requests and issues seemed to come over with only one problem, all issues were created by me. Not a horrible experience, but could be better (and maybe if the original creators had gitlab accounts, they would have been correctly assigned).