Ability to create a snapshot of current plugin versions and be able to restore from such a snapshot

[zsh-users]

Ability to create a snapshot of current plugin versions and be able to restore from such a snapshot

[Tarrasch]

I have for some time thought on something just like this. I like the names antigen-snapshot and antigen-restore. I definitely think this is a good next step for antigen, since what I like with pathogen (compared to vundle) is that with pathogen+gitmodules you commit your submodule hashes when you know that it will work because you have used it. I haven't tried Vundle yet, but afaik it's like antigen now. You don't have information about the versions of the submodules across your dotfiles repository.

An idea I had was to let antigen-snapshot not take any arguments. It could create something like the .gitmodules file and let antigen-snapshot read from that standard file and checkout those revisions. But there is really no "default" place to put such a file (I guess everyone doesn't have a dotfiles repository). So yea, I guess one should do as you said, that is to just provide the desired snapshot path as argument. Then the user can then decide weather they wanna check it in or just put it in a gist somewhere.

So yea, I agree with everything you said, it makes sense and I like the naming.

[GUIpsp]

Easiest way to do this IMHO is a newline separated pair of the location of origin and the hash

[sharat87]

@Tarrasch, you're right. Let's seeif Vundle ends up taking inspiration from antigen on this feature, that would be so much win :)

I haven't thought of the arguments to those commands, but come to think of it, what you say sounds good.

@GUIpsp, true. That is how the antigen-revert command is currently implemented. It should be fairly easy to bring that in to these commands.

[Tarrasch]

As for antigen-revert, I think antigen-revert could be a one-liner provided we have antigen-{snapshot,restore}. We can let antigen-update just internally do a antigen-snapshot on some file and let antigen-revert be an alias for antigen-restore that-file. Would this work? If so it could save us some code duplication.

[sharat87]

Yep, that's how it will be done.

[sharat87]

Okay, I just pushed a quick implementation and I got to go now. There is also a test which is very lacking, IMO. Please check it out and let me know what you think.

I still have to change antigen-update and antigen-revert to use these new commands.