Web Development Use Case

[StandardMachine]

I am curious about your thoughts on the future of tabular diffs. Currently as a LAMP developer, if you use git to manage your code base you have to resort to some extra tricks to get your DB changes turned into code and back again when moving from development to production. For example, in Drupal there is a system called Features to help do this.

However, one could imagine a scenario where someone could run both types of diffs and push them directly. Do you think that would that work and do you have a plan to make this thing the git of data?

[paulfitz]

Interesting question! The scenario you imagine is already sort of emerging. For example, jekyll-based github pages can include a _data directory with tables in it (encoded in yaml, json, or - very recently - csv). So a developer can edit a table that drives the site, and the templates that render the site, and push all that via git. Github has been busy adding good support for viewing csvs, and there are plugins to help out. So we can have pull requests against the code and data of sites. In this case, simple static sites, but there's no reason this should be limited to static sites.

I personally use daff and coopy all the time in web development for the Data Commons Co-op. Nothing very systematic. I think there's definitely a need and an appetite for what you envisage. For "small" data, decent workflows are starting to coalesce. For "big" data, you probably want something non-git-like optimized for the purpose. I'm hoping dat becomes that.