| commit | aedd85af7d7d0c90ed343ece4c66660db00841ec | |
| author | finnur@google.com <finnur@google.com@0039d316-1c4b-4281-b951-d872f2087c98> | |
| Thu, 4 Dec 2008 19:32:49 +0000 (4 19:32 +0000) | ||
| committer | finnur@google.com <finnur@google.com@0039d316-1c4b-4281-b951-d872f2087c98> | |
| Thu, 4 Dec 2008 19:32:49 +0000 (4 19:32 +0000) | ||
| tree | 8ba798b3932fbf2c1b5db69f39175326fd8745ab | treesnapshot (tar.gz zip) |
| parent | d5a94281a8df5b47a341e2dd71b4d5f8dafd5878 | commitdiff |
Fix issue 5079: Incorrect "Active match ordinal" count during Find-in-page
I introduced a regression in my reimplemenation of Find-in-page. The active match ordinal in Find-in-page (also known as "the 7" in "7 of 9") would be just a little off on pages with frames.
Problem A: When you search for something in gmail, for example, the ordinal could start off slightly negative or be 0. I wasn't checking the last_match_count_ of a frame for negative numbers before adding it to the total (it starts off as -1 and remains that way if the frame is not deemed to be worthy of being scoped, i.e. if it is hidden).
Problem B: On pages with multiple matches spread across multiple frames the ordinal would not be subtracted correctly after pressing F3 and Shift-F3 to go back to the frame you were on. We shouldn't be increasing/decreasing the active_match_index for a given frame when FindNext/FindPrevious causes us to jump between frames. We should instead reset it.
I added two tests to catch this in the future. They test ordinal values as you use Find in page (including combinations of frames/no-frames & FindNext/FindPrevious).
Oh, and I also removed some traces that were supposed to expose why a test was flaky, but it turns out to have been something unrelated to the test.
Review URL: http://codereview.chromium.org/13130
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@6369 0039d316-1c4b-4281-b951-d872f2087c98
I introduced a regression in my reimplemenation of Find-in-page. The active match ordinal in Find-in-page (also known as "the 7" in "7 of 9") would be just a little off on pages with frames.
Problem A: When you search for something in gmail, for example, the ordinal could start off slightly negative or be 0. I wasn't checking the last_match_count_ of a frame for negative numbers before adding it to the total (it starts off as -1 and remains that way if the frame is not deemed to be worthy of being scoped, i.e. if it is hidden).
Problem B: On pages with multiple matches spread across multiple frames the ordinal would not be subtracted correctly after pressing F3 and Shift-F3 to go back to the frame you were on. We shouldn't be increasing/decreasing the active_match_index for a given frame when FindNext/FindPrevious causes us to jump between frames. We should instead reset it.
I added two tests to catch this in the future. They test ordinal values as you use Find in page (including combinations of frames/no-frames & FindNext/FindPrevious).
Oh, and I also removed some traces that were supposed to expose why a test was flaky, but it turns out to have been something unrelated to the test.
Review URL: http://codereview.chromium.org/13130
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@6369 0039d316-1c4b-4281-b951-d872f2087c98