D2597: graft: add test for reading old graftstate files with new mechanism

[martinvonz (Martin von Zweigbergk)]

martinvonz added inline comments.

INLINE COMMENTS

> pulkit wrote in test-graft.t:1417
> That's a nice idea. I think we can, but the behavior will be bit different since graft will think the working directory now contains changes from `rev 1` and will straight away try to create a commit and later reporting that graft of 1 created no changes to commit. Here is the diff:
> 
>   ~/repo/hg-committed/tests$ hg diff
>   diff --git a/tests/test-graft.t b/tests/test-graft.t
>   --- a/tests/test-graft.t
>   +++ b/tests/test-graft.t
>   @@ -1414,16 +1414,9 @@ Testing the reading of old format grafts
>      $ hg add b
>      $ hg ci -m "bar to b"
>      created new head
>   -  $ hg oldgraft -r 1 -r 2 --config extensions.oldgraft=$TESTTMP/oldgraft.py
>   -  merging b
>   -  warning: conflicts while merging b! (edit, then use 'hg resolve --mark')
>   -  abort: unresolved conflicts
>   -  [255]
>   +  $ hg log -r 1 -r 2 -T '{node}\n' > .hg/graftstate
>    
>   -  $ echo foo > b
>   -  $ hg resolve -m
>   -  (no more unresolved files)
>   -  continue: hg graft --continue
>      $ hg graft --continue
>      grafting 1:80e6d2c47cfe "added b"
>   +  note: graft of 1:80e6d2c47cfe created no changes to commit
>      grafting 2:8be98ac1a569 "added c"
> 
> Since we are just testing reading old state format files, we should be fine with this. What do you think?

An alternative is to use the regular great command to start the process and then replace the .hg/graftstate once it's run into conflicts. That might be better so we have a repo state (include the merge state) that we know is valid to continue from.

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D2597

To: pulkit, #hg-reviewers
Cc: martinvonz, durin42, mercurial-devel