D3827: rebase: no need to store backup during dry-run while aborting

Yuya Nishihara yuya at tcha.org
Mon Jun 25 14:18:09 UTC 2018


>   > > - retcode = rbsrt._prepareabortorcontinue(abortf) +            # If in-memory, means aborting during dry-run, no need to backup +            backup = not rbsrt.inmemory +            retcode = rbsrt._prepareabortorcontinue(abortf, backup=backup)
>   >
>   > This seems confusing and is probably wrong since we wouldn't overwrite
>   >  `inmemory` to `False` if in-memory rebase were resumable.
>   >
>   > I think explicit `backup` flag is less bad.
>   
>   
>   @yuja Can I pass a indicator type flag to _origrebase() just to indicate that we are in dryrun and then I can use that flag to set values for `backup` and `suppwarns` (suppress warning when aborting rebase). I think it would be better than explicitly passing flags for each case. What do you say?

Sounds good.

A cleaner (but not simple) approach would be to stop calling `_origrebase()`
twice, and instead refactor `_origrebase()` and/or `rebaseruntime` to support
dry-run operation.



More information about the Mercurial-devel mailing list