skshetry

Thanks @eric-seppanen, @dtrifiro. I did not notice that. I'm lifting the https://github.com/iterative/dvc/labels/p0-critical.

This is not a bug. We cannot use stage cache to know if some stages will be re-run or not, we can only know so when we try to run...

@courentin, it looks like you have outputs duplication or some issue with graph correctness. Are you able to do `dvc dag` successfully?

@courentin, yeah it's wrong that it expects only one pipeline. https://github.com/iterative/dvc/blob/353e4cf3746345ee94a2d907d16c688e97fb1bfd/dvc/repo/graph.py#L41-L45 We should just do something similar to the following instead: ```python return chain.from_iterable(nx.dfs_postorder_nodes(graph, source=stage) for stage in stages) ```

> with odb caching we could probably also avoid spawning new odb's. Instantiating ODB does not involve anything other than simple `__init__`, and the underlying filesystems are already cached. I...

> > Instantiating ODB does not involve anything other than simple **init**, and the underlying filesystems are already cached. I did not expect that to be slow. > > Is...

I am fine with introducing the concept of partial results in `dvc data status --json` and exit with `2`, but I am not sure if we need to do this...

We can start with a simple way here, skip the broken file in the index, and assume that they are not broken in the HEAD. Since we are comparing between...

@dacbd, thank you for the reply. I am have one more question, why does it create a new PR with the same commit instead of overwriting with a new commit...

> but the original example was simplistic on purpose to clarify how it works. I disagree that it was simplistic because of all the reading/writing stuff it is doing, which...