103: Copy over comments on unchanged lines when updating the diff

alan.f******@gmai***** (Google Code) (Is this you? Claim this profile.)
chipx86
chipx86
Nov. 23, 2007
What steps will reproduce the problem?
1. Post a review request
2. publish a review with comments on some lines of code
3. upload a new diff (in my case using post-review --diff-only)

What is the expected output? What do you see instead?
I expect that the lines of code that have comments, and haven't changed
since the last diff should still have the comments attached.

Instead all the comment markers in the diff view disappear.

What operating system are you using? What browser?
Linux, Firefox 2

Please provide any additional information below.
This is probably an enhancement. Obviously, its more work to figure out
what comments can stay and which ones would have to leave, but without this
 we need to decide if its worth wiping out the comments to post a new diff.
david
#1 david
Maybe we can figure out how to figure out which comments still apply.  That sounds
like a hard problem, though.

This will probably be mitigated a bit once we have some simple UI to look at old
diffs.  The comments are all still available on the review request page, too.
  • -Type-Defect
    +Type-Enhancement
david
#2 david
  • +Copy over comments on unchanged lines when updating the diff
david
#3 david
  • +Component-DiffViewer
#4 alan.f******@gmai***** (Google Code) (Is this you? Claim this profile.)
You could use the interdiff tool to generate the incremental diff between the two,
and comments that apply to lines not in the incremental diff can probably be copied
over safely, and comments on lines in the incremental diff can be arked obsolete and
copied over.
chipx86
#5 chipx86
It's possible, yes, but still potentially problematic.

I have a change about to go up for review that will include comments in older
revisions of the diff in the same review. If the line didn't change, we can just keep
it on the old revision because it doesn't matter. If the line did change, the user
has to move it themselves anyway. The end result should be satisfactory for this bug.
  • +chipx86
chipx86
#6 chipx86
  • +Fixed