Support importing non-github git repositories about packagist

[xaav]

Is system() allowed on your host?

[Seldaek]

I have a dedicated server. But I thought of using https://github.com/patrikf/glip - if it's possible to read directly from a remote without cloning it that is. Probably it's not possible though. @stof you mentioned s2b clones git repos I believe? How do you handle this?

[xaav]

I used that library, and It has lots of bugs. I have created a bundle from that library (http://github.com/xaav/GitBundle), which has some fixes and improvements, but it is no where near a stable version. The reason for this is that it is very picky about what it can an cannot do (and believe me, cloning is much too complex for it). You may be able to make it work with a lot of commits and fixes, but you will have a lot of debugging to do. I really would like to improve this library to make it usable, but I don't have the time right now, you're much better off using https://github.com/ornicar/php-git-repo.

[stof]

@Seldaek we use ornicar's library, which uses git installed on the server.

[Seldaek]

I wasn't suggesting you clone the repo with it, you only need to access branches, tags, and composer.json files on those. Commit info would be nice too.

[Seldaek]

@stof yup, but it means you have to clone the repo right? It'd be nice to avoid that, for efficiency and avoiding a mess on the filesystem, but I guess most of the repos will be github anyway, so it shouldn't be too problematic.

[xaav]

@stof yup, but it means you have to clone the repo right? It'd be nice to avoid that, for efficiency and avoiding a mess on the filesystem, but I guess most of the repos will be github anyway, so it shouldn't be too problematic.

@Seldaek how do you plan on accessing the repo's data without cloning it? Even if there was a way to access the data without cloning the repository, all you would be doing is reading the data, and then throwing it away.

[Seldaek]

Well I was just curious if it was possible at all. It appears not though :)

[stof]

@Seldaek we indeed clone them. This allow us to retrieve some data from the code without the limitation of the github api (the v2 limits to 60 calls per minute, the v3 which we don't use currently limits to 5000 calls per day). For instance to get the readme (this could also be used to render the doc in the future)

[xaav]

It'd be nice to avoid that, for efficiency and avoiding a mess on the filesystem,

Just store the files in /tmp, then set a cron to clean out that folder.

[Seldaek]

Yup, it's not a big issue, just would have been nice to avoid it. Let's go with cloning & php git repo then. But this is not too urgent. Did you see I put stuff in milestones btw? :)

[Seldaek]

This could be more useful than phpGitRepo, remains to be seen.. https://github.com/teqneers/PHP-Stream-Wrapper-for-Git

[xaav]

Yes, but we would still need to build a zip of the repo.