Summary:
When a user hits 'cancel' on a 'new', 'edit', or 'reply' operation, add a little
"Changes discarded. __Undo__" insert so they can get their change back. No undo
for delete since there's an explicit prompt. Once this lands we can make
'escape' work again to close dialogs.
This change started feeling really good when I was merging all the duplicate
code and making things more consistent, but by the time I started writing client
rendering it felt gross. I'm not really thrilled with it but I guess it's a step
forward? The feature seems pretty OK in practice. Let me know how much barfing
this causes and I can try to remedy the most acute concerns.
This also fixes a bug where replies always (?) appear on the 'new' side of the
diff (I think?).
Test Plan:
Applied 'new', 'edit', 'delete' and 'reply' operations, pressed 'cancel' and
'okay' in each case, with and without changing text where relevant. All
behaviors seem to conform with expectations, except that canceling out of 'edit'
without changing the text gives you an option to undo when it shouldn't really.
There's no super easy way to get at the original text right now.
Reviewed By: aran
Reviewers: aran, jungejason, tuomaspelkonen
CC: simpkins, aran, epriestley
Differential Revision: 406
Summary:
Historically, we had a bug at some point which caused inline comments to get
associated with changeset 0. Prevent that explicitly. See T108.
Test Plan:
Set "$changeset = 0" in the endpoint and got an exception.
Reviewed By: aran
Reviewers: aran
CC: aran
Differential Revision: 374
Summary:
Fixes the issue caused by rPa0af5b66437719dba6136579c051982ab275e6a0. Prior to
that patch, isCommentInNewFile() returned $comment->getIsNewFile(). While this
was often the wrong value, it came from the database and was the integer 1 if
true.
After the patch, the function returns 'true' as a boolean, which is passed to JS
and then back to PHP, interpreted as an integer, and evaluates to 0.
To avoid this issue in general, provide an isBool() method on AphrontRequest
which interprets this correctly.
I will also revert the revert of rPa0af5b66437719dba6136579c051982ab275e6a0 when
I land this.
Test Plan:
Clicked "reply" on the right hand side of a diff, got a right-hand-side inline
comment.
Reviewed By: rm
Reviewers: tuomaspelkonen, jungejason, aran, rm
CC: simpkins, aran, epriestley, rm
Differential Revision: 250
Summary:
Thanks to erling for the report. This was XSSable, although you could
only get yourself.
Test Plan:
Made a comment like "</textarea><h1>" and edited it before and after
the patch. Proper behavior with this patch.
Reviewed By: aran
Reviewers: erling, jungejason, tuomaspelkonen, aran
CC: aran
Differential Revision: 187
Summary:
Currently if I go to leave an inline comment and then decide
against it and hit <tab>-<enter>, the inline comment edit dialog closes
but the overlay doesn't disappear due to a JS error around trying to
access the markup property of null. This fixes that by setting the
payload to an array with empty markup so that the JS works and the
overlay disappears.
Test Plan:
clicked on the line number column, then pressed ok and
confirmed the dialog closed, the overlay disappeared and there were no
JS errors. also tested that for clicking the reply link and when editing
an existing inline comment.
Reviewed By: aran
Reviewers: epriestley, aran
CC: aran, bh
Revert Plan:
OK
Differential Revision: 154
Summary:
Replies to comments went always to the left side of the diff. There
was a confusion between 'is_new' and 'on_right'. Database wants the
information if the reply is on the left or on the right. The database
does not care if the comment is a reply or an original comment.
The code looks a bit confusing, because the database field is called
'isNewFile' and that is used to determine, which side the comment
applies to.
Test Plan:
Tested that every of combination of new comment/reply to the left/right side
worked when editing and after submitting the comment.
Reviewed By: jungejason
Reviewers: jungejason
CC: epriestley, jungejason
Differential Revision: 148
