STR: 1. Open a comment/reply on an existing set of comments 2. Write lots of stuff (but don't save/submit) 3. scroll down to look at other code (you can now no longer see your written stuff) 4. notice an orthogonal issue, click on another comment marker to write a comment there ER: both comment boxes stay open, or my comment at least gets saved somehow/somewhere AR: everything I wrote is wiped immediately. :'-( This is very painful. I now have to go back and remember everything I wrote and hopefully re-write it more or less accurately, which is a lot of work. Mozreview shouldn't throw my stuff away so lightly.

Now there seems to be a prompt when I click cancel on a comment saying "are you sure", which is weird because I'm explicitly clicking 'cancel', so yes, I'm always sure. But still nothing if we implicitly click cancel by opening a different comment. Can someone explain how my usage scenario is so different from that of other people? I'm really confused why this was implemented the way it was...

Yeah that is really weird. I don't know why Review Board defaults to "you probably don't want this comment". We can fix this, though, and ensure the draft is saved when you click away. We did this recently with general review comments (bug 1291628).

Comment on attachment 8793651 [details] clone for mozreview modification (bug 1275865) https://reviewboard.mozilla.org/r/80372/#review79718

Comment on attachment 8793652 [details] Show confirmation when creating a new comment (bug 1275865) https://reviewboard.mozilla.org/r/80374/#review79724 One oddity I'm seeing is this: 1. Click on a line to make a comment. 2. Type some stuff without submitting 3. Click on another line 4. See your confirmation pop up, click "Cancel" 5. Nothing will happen to the current comment but the line you click will save have a "1" bubble on it. I'm going to R+ this anyway, since it's worth having this safeguard, but we should follow up to preven that "1" bubble from staying there.

smacleod: please land this if you think the issue raised in comment 6 isn't severe enough to warrant delaying this (fwiw I think this fix is better than the current behaviour).

i've fixed the issue david found in review in https://reviewboard.mozilla.org/r/80374/diff/1-2/; clearing ni.

Pushed by bjones@mozilla.com: https://hg.mozilla.org/webtools/reviewboard/rev/c1e10bd100f8 clone for mozreview modification r=davidwalsh https://hg.mozilla.org/webtools/reviewboard/rev/0a99e7ad3665 Show confirmation when creating a new comment r=davidwalsh