I have to handle quite a lot of Excel document at work, much to my despair. Unfortunately we are not using GIT, but SVN.
I ended up creating a small tool that extract the vba, named ranges plus the sheet in this format: "cell_adress,type,format,value"
I have to commit those files alongside the spreadsheet, which is not great but had helped me a lot.
Nice to hear others are working on the problem!
I have to say, pretty neat. And I partly feel like I may need a shower after thinking that.<p>I have no doubt that a product like this has a <i>real</i> need ... I used to write software targetted at end-users for reducing first-level support at a huge company, and it never ceased to amaze me the kinds of unbelievably important things that Excel and Access were used to manage (on the order of processes that the failure of which would result in 7-figure problems). It would further amaze me to find out that many of the people working with these end-user created solutions to problems were had some really, durably, solid rocks between their ears, and yet were still able to navigate these tools with <i>moderate</i> success. Unfortunately, there's little chance that an average excel hacker is going to have much interest or ability to operate git, and the average git enthusiast is probably a lot like me -- a guy who goes out of his way to <i>never</i> launch Excel for any reason, ever.<p>The last time I was given a task in which the work product was demanded in Excel, I wrote an application to create the spreadsheet programmatically ... partly because it was the only way to do some of the "clean up" they were asking for[0] and partly because I am a smart-ass by nature.<p>I realize I'm a <i>specific case</i> and I probably have a much stronger allergy to Excel than most, and as I said, there's few large companies that don't have stupidly-complex processes that (also stupidly-)rely on Excel. I'm also a guy who bends <i>git</i> into just about every scenario I can find, so I love the idea of anything that makes it more useful. Kudos.<p>[0] I had to correlate output from several web services and the source was a spreadsheet had the values they responded with. I wrote an app to read the rows and apply edit-distance against them to correlate a few more than were possible using just excel tricks.
Nice attempt, I looked into this problem too a while ago. Quick question for the OP, I know a lot of enterprise users want to version control their spreadsheets rather than just the code. Have you looked into that? Seems to be a tricky problem to solve.
Can anyone here answer whether it’s normal, when building git based interfaces, to only use the API for queries and contextual things but utilize regular git commands to perform interactions with the remote repositories?<p>I am a 3D artist building a system to version control the configuration/script file directory trees for all the applications I use on a day to day basis. I will be taking a look at this code for clues but maybe someone in this thread can help me with some questions.<p>Does anyone know something similar to this? I am modeling it on the VSCode Setting Sync extension, which is simple but a little rough itself.<p>I am using Python and just about have what I’m going for but it’s not as tight as I want it to be. I got a good system of using one repository with a branch per directory tree and folder symlinks to organize it locally. I can pull the branches into individual local repos per remote branch and it works nicely. This allows me to use regular git commands when I need to without fussing with other application’s configuration trees.<p>But things get pretty loose when it comes to building the backup() and restore() functions. For restore(), I can just doing an overwrite with fetch then reset hard —HEAD or various things like this but it seems like there must be a better approach considering the local repository, index etc. are a bit unnecessary. If anyone has any thoughts, links or pointers, I’d be super thankful! Sorry for the long post.
Hmm... I literally just created an AWS bucket that magically creates (i.e. lambda) corresponding .json files for any .xlsx file that gets put there.<p>I may have used Git if I knew this existed. For now our requirement was that the file be maintained in an Excel format, but this data needs to be consumed by different scripts written in different languages hence the automatic .json conversion.