Main differences between this fork and jedbrown version?

[justinclift]

Confusingly, this fork of git-fat and the original author version both seem to be under (separate) active development.

What are the advantages of this one vs the jedbrown one, and would you consider doing pull requests of your changes here, to merge back into one project? 😄

Just trying to reduce my confusion... 😉

[abraithwaite]

We cleaned up a lot and made it easier to change the code mostly, but we also added the ability to plug different backends (besides rsync) and also sped up the operation some.

On the pull request thread I opened, I was trying to get details on how to get it merged but was met with silence. 😦 I'll bump it again.

Really though, there aren't too many user visible changes besides the two I mentioned. It's pretty much the same software and should be interchangeable for the most part (I haven't done extensive testing on computability, but we didn't change the cookie format which is the biggest determining factor.)

I'll poke him again to see what's up.

[abraithwaite]

@justinclift Were the problems you're experiencing related to the retroactive import as well? From #52 on Jed's repo?

I'll investigate and see what I can find.

[abraithwaite]

Here's a hasty patch I just wrote for it. If you want to give it a shot then be sure to backup the repo before trying to use it.

https://gist.github.com/abraithwaite/95b6479871c38e6b30af

I manually tested it on my machine and it seems to work fine. A more extensive fix should come later with tests.

[justinclift]

@abraithwaite For me, I'm only starting to look at git-fat, and trying to work out which fork would best meet requirements. The flexible backend (eg more than just rsync) is probably going to win, but not if that fork is too buggy. 😉

[abraithwaite]

Ah okay. In that case I'm not sure which one has more bugs, but one of the reasons we refactored was so that we could make testing easier.

The retroactive import stuff was experimental and wasn't used by us so it's no surprise that there were bugs hiding in there. It's a pretty neat feature though so I'm going to clean it up and test it when I have time (perhaps this weekend?). I think it was also one of the things mentioned blocking Jed from merging my changes so that's an added bonus.

[justinclift]

I'll probably try the retroactive import stuff over this weekend. Wanting to convert/import a large (200GB+) old SVN repo full of media files to being git based, and trying to figure out the right approach. Will see how it goes, and open issues if appropriate. 😈

[abraithwaite]

Fixes are in on the develop branch if you want to try it out.

Cheers! 🍰

[justinclift]

@abraithwaite Your develop branch is good. Tried that, and both the jedbrown + your master branches to do a retroactive import on ~100GB git repo (full of media files, converted from Subversion). Only the develop branch seems to have made it to the end successfully.

Going to analyse it in more detail over the next few days, but so far this seems like a winner. 😄

[abraithwaite]

That's great news! I just got around to packaging it today, so now it should be installable by pip.

I'm glad it worked well for you! If you know of anything that might make importing easier, let me know. 🍰