Integrate VCS‌ as a first class part of the tool

[alerque]

I would like to propose adding some VCS integration on a first-class basis — not to make the font editor a full Git client of course, just to facilitate the bits where the font editor has more idea what is what than a third party generic VCS app would.

First detect whether the file being edited is also tracked in Git. If not, nothing changes. If so, add some options.

• A "commit" option along side save that would save, stage, and commit the relevant UFO files. An extra check should probably make sure the rest of the repository status is clean before enabling this and show the feature disabled if the cache has anything in it.

• Add a "revert" option that is basically an aggressive undo that backtracks to the last committed version. Perhaps this would function by stashing the current change, then resetting the file.

• Add comparison functions such as a glyph editor background that shows a greyed out outline of the last commited state of a glyph. The git object store can be queried directly for the last saved object.

More is certainly possible, but if VCS was brought in as a first class part of the tooling I think the overall workflow could be radically different (and better) than any other tooling that is out there.

If this is an agreeable proposition I'm even willing to help out a bit with some implementation work. I'm not much of a graphic designer but I do have a pretty deep understanding of how Git works and I'm just getting up to speed with Rust.

[eliheuer]

There is a Glyphs plugin that does something like this: https://github.com/simoncozens/GlyphsGit

It's nice to have depending on your workflow, I specifically like the idea of seeing old versions of a glyph from previous commits in the editor.

I wonder if this kind of thing would be better as a plugin, I'm not sure?

[alerque]

Apparently there are two similar Glyphs plugins because I actually saw this one: jenskutilek/Glyphs-SaveToGit.

[cmyr]

I think this is definitely an interesting feature idea, although I'm going to hold off on thinking about it too hard until we've got some more of the fundamentals in place. :)