I'm _mostly_ sure these are the only unsafe chunks here.
Details
- Reviewers
- None
- Group Reviewers
hg-reviewers - Commits
- rHGa1d5951efce7: narrow: mark the critical chunks of narrowing/widening as unsafe
Diff Detail
- Repository
- rHG Mercurial
- Lint
Automatic diff as part of commit; lint not applicable. - Unit
Automatic diff as part of commit; unit tests not applicable.
Event Timeline
hgext/narrow/narrowcommands.py | ||
---|---|---|
248 | Perhaps for another patch, but we could probably move this to before the transaction that starts on line 238 and then move the transaction outside of the unsafeoperation(). The repo won't be more broken if the user interrupts while we delete or add files from the working copy or dirstate here than it could be if they interrupt while doing the same things because of an hg update (right?). I guess it depends on what we consider unsafe. Is it just things that would result in lost commits or errors from hg verify that we consider unsafe? That seems like a reasonable definition to me. | |
250 | nit: the previous patch left this line outside of the block. make consistent? doesn't really matter, though... | |
284–296 | similar here: i think this could be left outside the unsafeoperation() |
Perhaps for another patch, but we could probably move this to before the transaction that starts on line 238 and then move the transaction outside of the unsafeoperation(). The repo won't be more broken if the user interrupts while we delete or add files from the working copy or dirstate here than it could be if they interrupt while doing the same things because of an hg update (right?).
I guess it depends on what we consider unsafe. Is it just things that would result in lost commits or errors from hg verify that we consider unsafe? That seems like a reasonable definition to me.