Reputation: 7227
We are developing a program that generates texts - aptly named the "text-generator". The texts change over time due to external requests.
We have scripts that generate representative sample texts, which we use as a kind of "test suite" to ensure that our code changes only have the desired effect and didn't introduce any bugs.
We are looking for a way to track how these generated sample texts change over time, ideally diffable. Checking these generated files into our git repository would work, but it bloats the commits, and reviewing difficult as code changes are mixed with generated-file changes. A separate repository would work, but where do you keep it, how do you ensure all code commits have a matching commit there, which commit from generated-repo corresponds to which code commit, ...
The ideal solution would be two trees associated to each git commit, once the source code tree, and then separately the generated-files tree, which would mostly be used for git diff
. But well, git can't do that.
Are there solutions out there that do what we want? It seems like we wouldn't be the first who want to keep track of how their code output changes.
Upvotes: 0
Views: 24
Reputation: 97282
Submodules (subtrees? not sure), if|when all samples are located in some common node - Git track and maintain relation between changesets in wrapper and nested object AFAICR
Upvotes: 1