User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:50.0) Gecko/20100101 Firefox/50.0 Build ID: 20160611030214 Steps to reproduce: Went to http://claws-mail.org/documentation.php?section=general and tried to find "pass" via Ctrl+f Actual results: "pass" appeared in a box with yellow background; but nowhere near that string in the page's text. Several other white borderless boxes also appear. Virtually impossible to find... Expected results: Find should align with the found text.

[Tracking Requested - why for this release]: Regression

The random white boxes and pass in the wrong position is a duplicate of bug 1279682. Only half the page is dimmed which is a duplicate of bug 1279681. Only about half the page is dimmed for me which makes it an extreme test case.

Bug 1279742 will fix these issues as well, because it's using a different method of building rectangles to cut out of the dimmed background.

This problem still exists...

(In reply to Mike de Boer [:mikedeboer] from comment #4) > This problem still exists... Does it? I can no longer replicate this.

I can reproduce the problem on Nightly50.0a1. https://hg.mozilla.org/mozilla-central/rev/c2da34d96746288b5fee27bf6542a12c9f410988 Mozilla/5.0 (Windows NT 10.0; WOW64; rv:50.0) Gecko/20100101 Firefox/50.0 ID:20160626030213

Feature has been backed out so not tracking these bugs for 50.

Bulk update to find bar bugs that won't be in 50 (according to https://bugzilla.mozilla.org/show_bug.cgi?id=1279695#c4 and local testing).

Comment on attachment 8786376 [details] Bug 1280149 - rework and fix the positioning for range rects to work better with iframes and try better to not draw rects that are not visible from the top document. This basically means to leverage the Geometry.jsm API more and better. https://reviewboard.mozilla.org/r/75344/#review73278 ::: toolkit/modules/FinderHighlighter.jsm:1069 (Diff revision 2) > if (dict.unconditionalRepaintRequested) { > dict.unconditionalRepaintRequested = false; > this._repaintHighlightAllMask(window); > - return; > - } > + // !unconditional. > + } else if (dict.modalHighlightRectsMap.size && pageContentChanged) { > - > - let { width, height } = this._getWindowDimensions(window); > - if (!dict.modalHighlightRectsMap.size || > - (Math.abs(dict.lastWindowDimensions.width - width) < kContentChangeThresholdPx && > - Math.abs(dict.lastWindowDimensions.height - height) < kContentChangeThresholdPx)) { > - return; > - } > - > - this.iterator.restart(this.finder); > - dict.lastWindowDimensions = { width, height }; > - this._repaintHighlightAllMask(window); > + this._repaintHighlightAllMask(window); > + } Can these two sections be merged? Can we always set dict.unconditionalRepaintRequested = false? if (dict.unconditionalRepaintRequested || (dict.modalHighlightRectsMap.size && pageContentChanged)) { dict.unconditionalRepatinRequested = false; this._repaintHighlightAllMask(window); } ::: toolkit/modules/FinderHighlighter.jsm:1072 (Diff revision 2) > + this.iterator.restart(this.finder); > + > if (dict.unconditionalRepaintRequested) { > dict.unconditionalRepaintRequested = false; > this._repaintHighlightAllMask(window); > - return; > + // !unconditional. This comment should be removed.

Pushed by mdeboer@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/cb40070e2bf6 rework and fix the positioning for range rects to work better with iframes and try better to not draw rects that are not visible from the top document. This basically means to leverage the Geometry.jsm API more and better. r=jaws

I cannot reproduce this issue because of bug 1305420 . This should be verified when bug 1305420 will be fixed.

Verified fixed using the the latest Nightly 52.0a1 (Build ID:20161005030211) on Windows 10, Ubuntu 14.04 and Mac OS X 10.10. What I observed is that the highlight for matching terms is visible outside the iframe area and moves when you scroll. For this issue I filled bug 1308187

Based on comment 6 from bug 1303248 setting status-51 disabled and based on comment 17 setting status-52 to verified