Branches not in metadata: branches out of range

[ben18785]

I have been really enjoying using targets for a project that me and a wider team of statisticians are on. We have now, however, run into this issue: whenever we run tar_make() or tar_load_globals() or tar_manifest() we get the following error:

Last error: branches not in metadata: branches out of range.

A google points us only to the original documentation for tar_read_raw.

I'm afraid I can't provide a reproducible example since our code is all unable to be shared (we're working on a government project) and I don't know what causes the issue so can't make a simpler version that reproduces the error.

But would someone be able to suggest where we might be able to start looking?

[wlandau]

Looks like you are using dynamic branching, either through the pattern argument of tar_target() or one of the batched replication functions in tarchetypes. It's hard to say exactly what's happening, but it looks like the expected dynamic branches went missing from the data store somehow. If you are working on a single project with a whole team, it is important to use separate per-person data stores. If multiple people call tar_make() on the same project at the same time, or there are multiple instances of tar_make() running at the same time on the same project, this can corrupt the data store and potentially cause the problem you mentioned. Otherwise, it's hard to say why it's happening in your case.

[ben18785]

Thanks. We aren't using dynamic branching actually (nor anything from tarchetypes), which is a bit strange. Is there a way to restart targets and just rerun everything?

[ben18785]

I've actually managed to locate the issue: someone on our team had put a tar_read statement inside one of the scripts. This was wreaking havoc with things. I wonder whether it might be possible to have a more informative error message?

[dewoller]

Thanks @ben18785 muchly, that clue pointed me in the exact right direction.

[billdenney]

Thanks for this discussion. I had this issue as well. My issue was that I was running tar_read() on a function instead of a target name.