On 3/21/2020 1:59 PM, Elijah Newren via GitGitGadget wrote: > This series provides a replacement for the sparsity updating in > sparse-checkout that is based on the logic from git read-tree -mu HEAD. The > most important bit is patch 9 and its lengthy commit message explaining the > current state and rationale for most the series, though patches 16 and 17 > have additional related directions and rationale for the series. Those three > patches are the most important to review. > > Changes since v1: > > * addressed several cleanups highlighted by Stolee (and I picked 'reapply' > for the new subcommand name) > * added three new minor cleanup commits (patches 2-4) > * several new patches to adjust the warning messages to look like warnings > (patches 11-15) > * display warning messages when paths are unmerged (patch 16) > * also make checkout and other unpack_trees()-using commands not error out > when unable to set the SKIP_WORKTREE bit (patch 17) I'm very happy with the updates in this version. > Also, the loop range in display_warning_msgs() in patch 14 might seem > slightly weird, but I felt adding another unpack_rejects array of some sort > and plumbing through an extra bit everywhere to notify the system which > unpack_rejects array to use would have been more invasive. Opinions either > way (or alternative approaches) welcome. > > I was also curious if I should have squashed all the patches to adjust the > warning messages (patches 11-15, they adjust the messages to actually look > like warnings instead of errors), since they are individually kind of small, > but it looked like it'd be much easier to review if I were to split them > apart, so I did so. I found them very easy to review. I had the one comment that was my fault for not looking one patch ahead. In general, it was simple to see that you were doing good updates with good motivations. This is going to help the UX around sparse-checkout quite a bit! Reviewed-by: Derrick Stolee <dstolee@xxxxxxxxxxxxx> -Stolee
