libgit2 VS immer

Everything that is replicated on the network is stored as a Git object, using the libgit2[0] library. This library uses hardened SHA-1 internally, which is called sha1dc (for "detect collision").

[0]: https://github.com/libgit2/libgit2/blob/ac0f2245510f6c75db1b...

This is delightful and I can't wait to try it out. Right now, the libgit2 project (https://github.com/libgit2/libgit2) has a custom HTTP git server wrapper that will throttle the responses down to a very slow rate. It's fun watching a `git clone` running over 2400 baud modem speeds, but it's actually been incredibly helpful for testing timeouts, odd buffering problems, and other things that crop up in weird network environments.

I'd love to jettison our hacky custom code and use something off-the-shelf instead.

I've built a UI on top of libgit2 and I wish that this blog post expanded on which new features are missing (sparse checkout?).

To quote: "The migration to Git Executable will allow us to resolve long-standing issues with GitKraken Client, such as poor LFS performance, SSH configuration support and many other features/performance improvements."

I agree on LFS performance on Windows. SSH config support is a pain due to libssh2 but openssh support is on the way (https://github.com/libgit2/libgit2/pull/6617).

There are many cons to using the Git executable itself (parsing output, error reporting, version handling). Seems to me that there's more to this?

I'm curious about the license change? This is an executable is it not? Invoking it as a separate process does not require you make the software calling it GPL so switching to MIT should have no affect in the common case.

If the authors really wanted a more permissive license, then instead of relicensing from AGPL to MIT they should have gone AGPL with linking exception. An example of a project that does this is libgit2 [1]. This licensing is more permissive but still permits the author to sell commercial licenses to those making closed-source code changes.

[1] https://github.com/libgit2/libgit2#license

Using libxlsxwriter and libgit, it's straightforward -- just putting the equivalent of git shortlog and lines added and removed into a line of cells.

I've been working on an editor (not text) in C++ and pretty early got into undo/redo. I went down the route of doIt/undoIt for commands but that quickly got old. There was both the extra work needed to implement undo separately for every operation, but also the nagging feeling that the undo operation for some operation wasn't implemented correctly.

In the end, I switched to representing the entire document state using persistent data structures (using the immer library). This vastly simplified things and implementing undo/redo becomes absolutely trivial when using persistent data structures. It's probably not something that is suitable for all domains, but worth checking out.

https://github.com/arximboldi/immer

How does this compare to https://github.com/arximboldi/immer (other than the C/C++ difference)?

Also, it's my understanding that, in practice, persistent data structures require a garbage collector in order to handle deallocation when used in a general-purpose way. How does your implementation handle that?

You might be interested in ewig and immer by Juan Pedro Bolivar Puente:

https://github.com/arximboldi/ewig

https://github.com/arximboldi/immer

See the author instantly opening a ~1GB text file with async loading, paging through, copying/pasting, and undoing/redoing in their prototype “ewig” text editor about 27 minutes into their talk here:

https://m.youtube.com/watch?v=sPhpelUfu8Q

It’s backed by a “vector of vectors” data structure called a relaxed radix balanced tree:

https://infoscience.epfl.ch/record/169879/files/RMTrees.pdf

That original paper has seen lots of attention and attempts at performance improvements, such as:

https://hypirion.com/musings/thesis

https://github.com/hyPiRion/c-rrb

You’re absolutely right, however people have been putting in the “extra efforts” required for efficiency. Check out immer if you’re interested.

The C++ immer library: https://github.com/arximboldi/immer

For C++ check this one out - https://github.com/arximboldi/immer

Yeah I agree that I should link that wikipedia page in the docs, I'll do that as soon as I get a chance. immer (https://github.com/arximboldi/immer) also links that page in its docs, for the exact same reason I'm sure. Interestingly, there is a lot of overlap between persistent data structures in the functional programming sense and persistent data structures in the persisted-to-disk sense because persistent data structures in the FP sense are one of the best ways to guarantee atomic updates and safe failure recovery in a persisted-to-disk system! Btrfs and ZFS, as well as many databases, are at their core basically just copy-on-write B+ trees.

I’ve never tried it, but I love the idea of implementing editor scene state using immutable data structures like https://github.com/arximboldi/immer With that, every edit would append a new node to a list of scene states. Undo/redo becomes iterating your view of the scene up and down through that list. Can’t screw up an undo function if there’s never any work to do :P

I have! I think it's related to the C++ immer library which I used several years ago in Vortex. It's kinda like the previous generation of ValueScript. 🍻