In D6731#98871, @pulkit wrote:

In such cases, I like the idea of having fix as two patches, first which demonstrates the bug and the second which fixes the bug. What do you think?

In D6479#98757, @navaneeth.suresh wrote:

In D6479#98737, @durin42 wrote:

The only thing I'm curious about really is why we have extrastorage and usualstorge. Can we get away with only one of those choices instead of having more options?

The ultimate goal is to store the mergestate info in the changeset extras only. But, now , it has no support to store the local version of the files which are stored in the mergestate. This lacks info when the user gets conflicts other than a usual hg merge.

In such cases, I like the idea of having fix as two patches, first which demonstrates the bug and the second which fixes the bug. What do you think?

In D6730#98852, @av6 wrote:

The changes in tests that this patch causes are actually already present in the D6709, so let's fold this patch into it as well.

The changes in tests that this patch causes are actually already present in the D6709, so let's fold this patch into it as well.