[question] Is git meta dying or just really stable?

[jhdub23]

There has been very little activity. Is git meta just that stable and will be used/supported by twosigma & others long-term, or is it slowly fading away? I'm looking to deploy git meta into our company, but not if this is a dead end.

Thanks,

Jay

[lordmauve]

Two Sigma continues to use git-meta in production, and it has been very successful. Our monorepo holds some 500GB of objects and users have effective workflows to work with that. In that sense it is stable.

However we are exploring the idea of using sparse checkouts of a unified repo instead of git meta. We don't know if this effort will be successful (git isn't scaling perfectly, GitLab is struggling in numerous places). So it is too early to say if git-meta has a long term role. In the meantime we have reduced our investment in git-meta. I wouldn't say it is in maintenance-only mode but we have not scheduled any development work on git-meta for the upcoming quarter.

[jhdub23]

Thank you for the feedback. I have been kicking the tires of git meta off and on and have been trying to sell the idea within our company for quite a while.

I have tried sparse checkouts and have also experienced scaling/performance issues. I'm assuming that this will only get better over time, but it's a question of timeframe. Also, we have many product groups who want to limit access to their repos for security reasons, and creating a unified repo is an even harder challenge. I'm at a decision point on whether or not to pull the trigger on a partial deployment of git meta between willing groups.

I very much appreciate your feedback. We will play follow the leader here and would be interested in knowing if you find a unified repo viable.

[mfarrugi]

@lordmauve how has this panned out?

[leifwalsh]

git meta continues to do what it does, which is provide a higher level interface to a specific workflow built around git submodules. It is stable and we continue to use it, but we haven't been able to keep pace with development of modern git features, partly due to limitations of libgit2 on which node-git is based, but more commonly due to lack of investment from our side.

If it does something valuable for you, that's great and you should enjoy it! But I'll say that internally, we've been following developments in sparse/partial clone improvements in upstream git and related projects and we're quite excited about them, and looking to possibly migrate to native git workflows where we would not require submodules or this project.

[martinRenou]

migrate to native git workflows where we would not require submodules or this project.

❤️