38704 2010-05-06 16:51:30 -0700 SVG pattern size changed when resizing browser 2010-08-19 04:55:43 -0700 1 1 1 Unclassified WebKit SVG 528+ (Nightly build) PC OS X 10.5 RESOLVED DUPLICATE 41396 HasReduction, InRadar P2 Normal --- 1 s3lance webkit-unassigned bdakin krit pkchan+bugzilla=webkit zimmermann oldest_to_newest 222060 0 s3lance 2010-05-06 16:51:30 -0700 * SUMMARY The pattern size changed when resizing browser, which the pattern size should be independent of the browser's size. The pattern even failed to render repeatedly if we resize the browser to smaller than the element which applied the pattern and then scroll down. The problem doesn't exist if we don't use viewBox, but it is needed. * STEPS TO REPRODUCE (see test-case in attachment) 1. Define a pattern with {x=0, y=0, width=700, height=700} 2. Set the pattern with viewBox {0 0 200 200} 3. Apply the pattern on a rect with {x=0, y=0, width=800 height=800} 4. Resize browser to smaller than the pattern size 5. Scroll down * EXPECTED RESULTS The pattern size doesn't change when resizing and scrolling * ACTUAL RESULTS pattern size changed in step 4 pattern failed to render repeatedly in step 5 * REGRESSION Safari Version 4.0.4 (6531.21.10, r53178) / Mac It seems that the size of SVG pattern is upper-bounded by the size of the browser window. 222062 1 55321 s3lance 2010-05-06 16:52:17 -0700 Created attachment 55321 Test case 222063 2 s3lance 2010-05-06 16:52:59 -0700 <rdar://problem/7541466> 222168 3 krit 2010-05-06 22:19:33 -0700 So the problem is, that the pattern is cut to the actual viewport, but doesn't rebuild if the size of the windows changes or if we scroll a bit. I looked at an older revision, and even the size of the pattern changes with resizing the window. That is wrong to. My proposal is to not clip to viewport anymore on creating a pattern in RenderSVGResourcePattern.cpp. 222171 4 krit 2010-05-06 22:31:25 -0700 (In reply to comment #3) > So the problem is, that the pattern is cut to the actual viewport, but doesn't > rebuild if the size of the windows changes or if we scroll a bit. > > I looked at an older revision, and even the size of the pattern changes with > resizing the window. That is wrong to. > > My proposal is to not clip to viewport anymore on creating a pattern in > RenderSVGResourcePattern.cpp. Clipping itself is realy dangerous. I thougt about at least clipping to the size of the target element, but we have to respect pattern transformations. While scale and translation isn't a problem, determin the visible size of a pattern on skewing or rotation is realy difficult. 222173 5 s3lance 2010-05-06 22:34:45 -0700 Yep. Just tried removing the call to clampImageBufferSizeToViewport(object->document()->view(), imageSize) it works :-) Although the repaint issue (if the browser client area size is smaller than the pattern filled rectangle and you scroll it) is still there. 222179 6 s3lance 2010-05-06 22:48:57 -0700 > While scale and translation isn't a problem, determine the visible size of a > pattern on skewing or rotation is realy difficult. We should be able to deal with the easy case first: check if the affine transform matrix to see if is in the form: [ a 0 e ] [ 0 d f ] [ 0 0 1 ] So that we know if there's no rotation and skewing. The decomposition of this particular case is trivial. 222180 7 s3lance 2010-05-06 22:56:13 -0700 > Clipping itself is realy dangerous. I thougt about at least clipping to the > size of the target element, but we have to respect pattern transformations. > While scale and translation isn't a problem, determin the visible size of a > pattern on skewing or rotation is realy difficult. By the way, do you think it's possible to simply use GraphicsContext::setFillPattern() / GraphicsContext::applyFillPattern() instead of implementing it ourselves?? 222181 8 krit 2010-05-06 23:02:59 -0700 (In reply to comment #7) > > Clipping itself is realy dangerous. I thougt about at least clipping to the > > size of the target element, but we have to respect pattern transformations. > > While scale and translation isn't a problem, determin the visible size of a > > pattern on skewing or rotation is realy difficult. > > By the way, do you think it's possible to simply use > GraphicsContext::setFillPattern() / GraphicsContext::applyFillPattern() instead > of implementing it ourselves?? Not sure what you mean? We use context->setFillPattern(patternData->pattern); and context->fillPath(). Simiular to the HTML Canvas code. 222186 9 s3lance 2010-05-06 23:07:18 -0700 > Not sure what you mean? We use context->setFillPattern(patternData->pattern); > and context->fillPath(). Simiular to the HTML Canvas code. oops ... please ignore that :) I'm looking at RenderSVGResourcePattern::buildPattern newTileImageContext->save(); newTileImageContext->translate(-patternBoundaries.width() * numX, -patternBoundaries.height() * numY); for (int i = numY; i > 0; --i) { newTileImageContext->translate(0, patternBoundaries.height()); for (int j = numX; j > 0; --j) { newTileImageContext->translate(patternBoundaries.width(), 0); newTileImageContext->drawImage(tileImage->image(), style()->colorSpace(), tileRect, tileRect); } newTileImageContext->translate(-patternBoundaries.width() * numX, 0); } newTileImageContext->restore(); and i saw a method called DrawTiledImage() in GraphicsContext I just want to know if it could do the same thing 222189 10 krit 2010-05-06 23:10:27 -0700 (In reply to comment #6) > > While scale and translation isn't a problem, determine the visible size of a > > pattern on skewing or rotation is realy difficult. > > We should be able to deal with the easy case first: check if the affine > transform matrix to see if is in the form: > [ a 0 e ] > [ 0 d f ] > [ 0 0 1 ] > So that we know if there's no rotation and skewing. The decomposition of this > particular case is trivial. We have isIdentityOrTranslationOrFlipped() you can start with. If we want to clip to the target element boundaries, we need to be carefull, if we fil the target or stroke the target with the pattern. It's saver to take the strokeBoundaries of the object instead of the boundingBox. 222190 11 krit 2010-05-06 23:14:28 -0700 (In reply to comment #9) > > Not sure what you mean? We use context->setFillPattern(patternData->pattern); > > and context->fillPath(). Simiular to the HTML Canvas code. > > oops ... please ignore that :) > I'm looking at RenderSVGResourcePattern::buildPattern > > newTileImageContext->save(); > newTileImageContext->translate(-patternBoundaries.width() * numX, > -patternBoundaries.height() * numY); > for (int i = numY; i > 0; --i) { > newTileImageContext->translate(0, patternBoundaries.height()); > for (int j = numX; j > 0; --j) { > newTileImageContext->translate(patternBoundaries.width(), 0); > newTileImageContext->drawImage(tileImage->image(), > style()->colorSpace(), tileRect, tileRect); > } > newTileImageContext->translate(-patternBoundaries.width() * numX, 0); > } > newTileImageContext->restore(); > > > and i saw a method called DrawTiledImage() in GraphicsContext > > I just want to know if it could do the same thing it's a rarely used case and was needed by css-property overflow. The old code had a switch and we didn't run throug that code the most of the time. I need to look at the history, why this switch was removed ... 222192 12 krit 2010-05-06 23:17:44 -0700 (In reply to comment #11) > (In reply to comment #9) > > > Not sure what you mean? We use context->setFillPattern(patternData->pattern); > > > and context->fillPath(). Simiular to the HTML Canvas code. > > > > oops ... please ignore that :) > > I'm looking at RenderSVGResourcePattern::buildPattern > > > > newTileImageContext->save(); > > newTileImageContext->translate(-patternBoundaries.width() * numX, > > -patternBoundaries.height() * numY); > > for (int i = numY; i > 0; --i) { > > newTileImageContext->translate(0, patternBoundaries.height()); > > for (int j = numX; j > 0; --j) { > > newTileImageContext->translate(patternBoundaries.width(), 0); > > newTileImageContext->drawImage(tileImage->image(), > > style()->colorSpace(), tileRect, tileRect); > > } > > newTileImageContext->translate(-patternBoundaries.width() * numX, 0); > > } > > newTileImageContext->restore(); > > > > > > and i saw a method called DrawTiledImage() in GraphicsContext > > > > I just want to know if it could do the same thing > > it's a rarely used case and was needed by css-property overflow. The old code > had a switch and we didn't run throug that code the most of the time. I need to > look at the history, why this switch was removed ... Ah, here is it: http://trac.webkit.org/browser/trunk/WebCore/rendering/RenderSVGResourcePattern.cpp#L291 222212 13 55345 s3lance 2010-05-07 00:13:40 -0700 Created attachment 55345 Patch (without optimization) 222219 14 s3lance 2010-05-07 00:35:36 -0700 > We have isIdentityOrTranslationOrFlipped() you can start with. If we want to > clip to the target element boundaries, we need to be carefull, if we fil the > target or stroke the target with the pattern. It's saver to take the > strokeBoundaries of the object instead of the boundingBox. We can also determine if the matrix represents any skewing too Consider the 2x2 portion of the CTM [ a b ] [ c d ] for rotation, scaling, and flipping, the row vectors [ a b ] and [ c d ] are orthogonal, thus we can check if the dot product is (close to) zero then, we could decompose the matrix by normalizing the vector [ a b ] and [ c d ], determine flipping by inspecting the signs of the components, and use atan2() to find the rotation angle perhaps skewing is hard because even if we know the skewed pattern tile rect fully encloses the clipping rect, we still have to think how we should clip it 222273 15 zimmermann 2010-05-07 04:47:36 -0700 I'm hestitated to r+ the patch, some background notes on the clipping. The intent was to avoid creating uber-large ImageBuffers that may consume all your memory. Can you investigate whether we can find a better way to handle this problem? What about LayoutTests/svg/custom/pattern-excessive-malloc.svg - doesn't it blow up with your patch? (aka. crash) 222570 16 55345 krit 2010-05-07 12:57:36 -0700 Comment on attachment 55345 Patch (without optimization) We should realy think about clipping here. I maybe mixed up this bug with another clipping bug we had with pattern. On the other bug, we also clipped the content of the pattern to viewPort, and that had the problems with determing the correct drawing are on skewing. Maybe this is not quite relevant here. We know the strokeBoundaries of the target-element, and the tmpImageBuffer for the pattern don't need to be bigger (after all transformations) than the strokeBoundaries of the target. We need some test cases, but maybe Niko is right and we don't care about the type of transformation, but just about the resulting size of the pattern. And this is quite easy to check. We can transform the current boundaries of the Pattern with the transfromation matrix of the pattern and clip it to the strokeBoundaries of the target. 222573 17 krit 2010-05-07 13:02:18 -0700 (In reply to comment #16) > (From update of attachment 55345 [details]) > We should realy think about clipping here. I maybe mixed up this bug with > another clipping bug we had with pattern. On the other bug, we also clipped the > content of the pattern to viewPort, and that had the problems with determing > the correct drawing are on skewing. > Maybe this is not quite relevant here. We know the strokeBoundaries of the > target-element, and the tmpImageBuffer for the pattern don't need to be bigger > (after all transformations) than the strokeBoundaries of the target. > We need some test cases, but maybe Niko is right and we don't care about the > type of transformation, but just about the resulting size of the pattern. > And this is quite easy to check. We can transform the current boundaries of the > Pattern with the transfromation matrix of the pattern and clip it to the > strokeBoundaries of the target. The other bug I mixed up this bug was bug 29911. 224163 18 55758 s3lance 2010-05-11 14:35:55 -0700 Created attachment 55758 Patch 224168 19 55760 krit 2010-05-11 14:48:50 -0700 Created attachment 55760 Test case SVG Test case. 224176 20 krit 2010-05-11 15:10:04 -0700 (In reply to comment #19) > Created an attachment (id=55760) [details] > Test case > > SVG Test case. Note, that pattern were faster and had better results before the code move to the Renderer. 224444 21 zimmermann 2010-05-12 00:10:41 -0700 (In reply to comment #20) > (In reply to comment #19) > > Created an attachment (id=55760) [details] [details] > > Test case > > > > SVG Test case. > > Note, that pattern were faster and had better results before the code move to the Renderer. Huh? I almost can't believe you on that one, please elaborate. If you are aware of problems, why didn't you file a bug? 224449 22 krit 2010-05-12 00:21:25 -0700 (In reply to comment #21) > (In reply to comment #20) > > (In reply to comment #19) > > > Created an attachment (id=55760) [details] [details] [details] > > > Test case > > > > > > SVG Test case. > > > > Note, that pattern were faster and had better results before the code move to the Renderer. > Huh? I almost can't believe you on that one, please elaborate. If you are aware of problems, why didn't you file a bug? I realised the perf problems with the test above and I created the test just for this bug. I think that the patch (that still needs to be uploaded ;-)) will catch the issues on the test as well, since it looks like a to big imageBuffer is used. But I'm not sure how we can cover this problem. The current pattern already has a size of 10000x10000. With the transformation of patternTransform (scale of 0.001) it gets 1000 times bigger. At the moment we clip this to the viewport size, so it is at least displayable. 241066 23 55758 abarth 2010-06-21 19:30:14 -0700 Comment on attachment 55758 Patch This patch might well be a good code change, but it doesn't have any tests. If this isn't possible to test in DRT (can we resize the viewport in DRT?), then please at least add a test case to manual-tests. 266650 24 zimmermann 2010-08-19 04:55:43 -0700 *** This bug has been marked as a duplicate of bug 41396 *** 55321 2010-05-06 16:52:17 -0700 2010-05-06 16:52:17 -0700 Test case resize_bug.zip application/zip 3831 s3lance UEsDBAoAAAAAAMKbLjwAAAAAAAAAAAAAAAALABAAcmVzaXplX2J1Zy9VWAwAMQBPSzsAT0v1ARQA UEsDBBQACAAIAFxXeTsAAAAAAAAAAAAAAAAYABAAcmVzaXplX2J1Zy9pbnNwZWN0b3IucG5nVVgM AEgAT0tvnQxL9QEUAAFsB5P4iVBORw0KGgoAAAANSUhEUgAAACAAAAAgCAYAAABzenr0AAAACXBI WXMAAAsTAAALEwEAmpwYAAABMmlDQ1BQaG90b3Nob3AgSUNDIHByb2ZpbGUAAHjarZG9SsNQGIaf 04qCQxAJbsLBQVzEn61j0pYiONQokmRrkkMVbXI4Of508ia8CAcXR0HvoOIgOHkJboI4ODgECU4i +EzP9w4vL3zQWPE6frcxB6PcmqDnyzCK5cwj0zQBYJCW2uv3twHyIlf8RMD7MwLgadXr+F3+xmyq jQU+gc1MlSmIdSA7s9qCuATc5EhbEFeAa/aCNog7wBlWPgGcpPIXwDFhFIN4BdxhGMXQAHCTyl3A tercArQLPTaHwwMrN1qtlvSyIlFyd1xaNSrlVp4WRhdmYFUGVPuq3Z7Wx0oGPZ//JYxiWdnbDgIQ C5M6q0lPzOn3D8TD73fdMb4HL4Cp2zrb/4DrNVhs1tnyEsxfwI3+AvOlUD7FY+VVAAAAIGNIUk0A AHolAACAgwAA9CUAAITRAABtXwAA6GwAADyLAAAbWIPnB3gAAAW0SURBVHja5FdZbFRVGP7O3Wfv zHQvlFJFrE1btSwuQUnUsseECDFK4gOJxqiJGn0wIYaQkBgbEtTIA4kIrpGmBpcKKiQYJDWWQkVB sGWo1ZLOdJnpzJ0799655xwfLCPSmWnBBx48yZ+cm/uf/N///cv5D+Gc40YuATd4SZc3hJCiis1t 7R5NxnJwYR0BqeVg9YDgAeEm4aSfgw+D8y5bFY/2HXghMZPhy8yT3KYAgKWr3vRzIfuc261tXthQ F2hsuslVXh6SgyF/DnxyMk2jI+P22TMR89ezkVQ6ZXSKgrCj+8sXh/8TgMWrX18pydKOJXc119yz rNVnc1kYTxEkdIBxwMwCmgyIAhD0cpT6AQlZ3vPD6VT3sVNRy85u6fnq5f3XBWDRiteeCZUGt6x+ pC2sekPy8LiAtElmjKnfzVEdZqCWTr/Yf2gsNjK6s95X397RsZHOGsCdK157prqm4tUH17eVJk23 MJEUrzm5ykoc+DSLHz7w7dilweE3TnzzyvZZAVjUtv2hQDi4b81jD1eMpVyCblx/oQS8DEFPhn35 0Wex+MTEs6e+3tJZFEBzW7tHJvb3ax9ffxuTy5R4qrjxOWERq1pdAICDvRn8OU6n6ZSWUIg04Xy+ ryOSNe2lfUe3Jq4E8C8LEjOerG+4uVbxhpTxSYAxVlTub1TQVCuhqVbCyjvUvDqxCQLZ5ZcWtjSW CxJ5vmAjWr58qwbgqaa7FwViEwBjdEbpi5gwLAbOObrPZQrqxeIcDUta/EQUn1jywCvhvI0oAWdl eXVNiBKXmMnQWcX41IUMzv1pAgAyVuGWrutAiV8RKmurvX8MXFwD4L1pDDDwe6vq5mlJnYGy2Qtj DEEPmVEvpXNU1dVq4HRZXgY4p82hqkrVMBgYLezN3FIZG5f5EfaLCPv+Ls+fLprY1RUvypZhEIQq K1XOWWPeHOBwKhSXS7QsB5TSgjIYNfHxd3FoMs8lWm+/UfQMpRSmRaG43RLntDI/AxSKIMnEcWaO /2iCQ5NJrpRODqTBGCt6hlICIiqEMyZv2LBBBECvDoFtmiYDZIHS4jNCc52WM9gXySBlZGe+9wUC y7IY54x1dHTQfDkQ1ZPJWwkJSowW96Zlvprzvrc/DUZnZk1QRaSTKUq4M5x/IGHsTCIWtWWleEar MtAyX8vF/8Rv+qyqRVYJ4qNRmzJ+Li8AQnlX9OIFS3OLRZvP7fV/e885x1DMgp7J4tH7ShDykqLn VJeIkciACY6uvEloeNLHY4MDCeoYQVEiYtbKT+stNf/Qf3bIwLPrymBYDLG4XZB+zS2BOSYbGThv mJZ+JC8Dg0f3moyxdy70dicDIaVg/+eM5xh46A4/0hmK3V3RondGIKTiYl9PynHs9wf79ibyMgAA TEjvGvq5d1PlgkaP2+dX9KQ1zZv3j8Sgmw4A4NgvKQzFrOIDSkiDmY5nIye7oza337r6/7R5YEHr 5jWy27V76YbNFck4E81ZlFih5fIqCIRE1tO5L2YkEi8NnNzz4dXzwLQLv7/3na5sWt/Zd/CTMV+J QN0+JUf5tYi3RIU/KLLThzrHjcTEniuNz/gu6D+5t31y5NLOnk/3xEQhZYeqvJCU2Y1lsioiXOOD LNtOT+ee2MTw4L4AUbYW0i86Fc9v2bRREIVt1Y2t5XWty/yMy6KVziJrOXBsCs4BIhBIsgBZk6B5 FAiCw4ZOHU/+cfrHBHfotshPH7x7PVOxC4APgBAoXTjHX9bwtOLyrS2rb3CF59+qugIhwRUIiaKs EE4p0vFRx0wm2Pjv563RgbNm1pg8nJwcfDt+qS8y1fPjALLXAqAcQB0AdWo/V1Z9t3n8cxarntJ5 ouT2SrImXyaROlbWcdIpOxMf0hNDPVlz8gyAYQBjADIAfgWQui4GpvbeqW//1N4NQJn67wCwABgA 9ClDOoA0AHPWDPxvX8d/DQDeN0N6VtWsMQAAAABJRU5ErkJgglBLBwiZojLrcQcAAGwHAABQSwME CgAAAAAAyJsuPAAAAAAAAAAAAAAAAAkAEABfX01BQ09TWC9VWAwASABPS0gAT0v1ARQAUEsDBAoA AAAAAMibLjwAAAAAAAAAAAAAAAAUABAAX19NQUNPU1gvcmVzaXplX2J1Zy9VWAwASABPS0gAT0v1 ARQAUEsDBBQACAAIAFxXeTsAAAAAAAAAAAAAAAAjABAAX19NQUNPU1gvcmVzaXplX2J1Zy8uX2lu c3BlY3Rvci5wbmdVWAwASABPS2+dDEv1ARQAjY/PSsNAEMZnA+Kfkxc95+JBD9k01EqSgwiieCiK LagnmWw3zWKyu92dHgQfwNfwETz7Aj6WKw20ePIHM8N8fAPfwNbhNkQAYxTx7SR+jHt+NdgNlQGw 6zDDzl7hX1xMp/dR9HXSX3yH+vxjiXr9HeBImC5Ba1uZdJJwhoTFy/jyhmT30Egnr5zpfLCyu9CO AQ7W/sUSHWpSWkJlW+UpTT/Y8/5TQ2QLzlsjsG2Mp2KYZin3hKQEt4pLzcXSOamJqw7n0nMypq3Q VUsioz1X2lspyLjE6vn5IDs9y7PRIM939lbxGdv4Y5N6wdNAOaxSkc9GdTnBGp0q39ahVwr8AFBL Bwil4tQyAAEAAHkBAABQSwMEFAAIAAgAjpsuPAAAAAAAAAAAAAAAABkAEAByZXNpemVfYnVnL3Jl c2l6ZV9idWcuc3ZnVVgMAEgAT0vc/05L9QEUAH1SwU7jMBC99ysGc6HSNnbgsNsqqQQtq0WiC4JU uz2GxkksHDuyTZP+/Y7ddOkBsOQ4Y888v3l+ydnyYZFtHm+hdo2Ex/XN/d0CyITSP1cLSpfZEv7+ ylb3EEcMnp0RW0fp7W8CpHaunVHadV3UXUXaVDR7or1HiX3Z8DuxoSYqXEHmoyRc0jdS2fQDgHg6 nR7qfO6LLva42F0FnShcnZIfjBGouahqNwTW7SVPid5xU0rdzfI3pwlgZIVWKYmjmHx+3SVjjCL8 kDLrpVCvn/Pyp8hrBDiSCpzJlS21aVJit7nkF/EYTyGMpOClPQYYtrlz3CgQRUp+CikJDDtrJRxy e7PcPLf5lj+oteXIJyXY3D58h96/n/YeggEie+cRKMncIRf2DWI2RikE7260xwMG2LCfWGo43rjj 17blW/eUO6FT0q+E2qzyHhrOHXlnj/xFk1f8I1aXp6y+hC42OA/QEKSc1YaXKREqJGoTtaoi9EQ0 OjQ4h6Oq9FTWxGAZlKgm6mfkxbkXdhy0iw80/fKVdYx+Re+cszCOG5OhIP6/ofFhhPN4R3oJrdAH ifcOGpQORqXeuPPRP1BLBwh8lbGeuQEAAFIDAABQSwMEFAAIAAgAjpsuPAAAAAAAAAAAAAAAACQA EABfX01BQ09TWC9yZXNpemVfYnVnLy5fcmVzaXplX2J1Zy5zdmdVWAwASABPS9z/Tkv1ARQAY2AV Y2dgYmDwTUxW8A9WiFCAApAYAycQGwFxJRCD+KsZiAKOISFBTEy7NaE65gAxP5oSRoS4eHJ+rl5i QUFOql5IakWJa15yfkpmXjpQsrQkTdfC2tDYxMjQ3NLCBABQSwcI4kQ0eF8AAACrAAAAUEsBAhUD CgAAAAAAwpsuPAAAAAAAAAAAAAAAAAsADAAAAAAAAAAAQO1BAAAAAHJlc2l6ZV9idWcvVVgIADEA T0s7AE9LUEsBAhUDFAAIAAgAXFd5O5miMutxBwAAbAcAABgADAAAAAAAAAAAQKSBOQAAAHJlc2l6 ZV9idWcvaW5zcGVjdG9yLnBuZ1VYCABIAE9Lb50MS1BLAQIVAwoAAAAAAMibLjwAAAAAAAAAAAAA AAAJAAwAAAAAAAAAAED9QQAIAABfX01BQ09TWC9VWAgASABPS0gAT0tQSwECFQMKAAAAAADImy48 AAAAAAAAAAAAAAAAFAAMAAAAAAAAAABA/UE3CAAAX19NQUNPU1gvcmVzaXplX2J1Zy9VWAgASABP S0gAT0tQSwECFQMUAAgACABcV3k7peLUMgABAAB5AQAAIwAMAAAAAAAAAABApIF5CAAAX19NQUNP U1gvcmVzaXplX2J1Zy8uX2luc3BlY3Rvci5wbmdVWAgASABPS2+dDEtQSwECFQMUAAgACACOmy48 fJWxnrkBAABSAwAAGQAMAAAAAAAAAABApIHaCQAAcmVzaXplX2J1Zy9yZXNpemVfYnVnLnN2Z1VY CABIAE9L3P9OS1BLAQIVAxQACAAIAI6bLjziRDR4XwAAAKsAAAAkAAwAAAAAAAAAAECkgeoLAABf X01BQ09TWC9yZXNpemVfYnVnLy5fcmVzaXplX2J1Zy5zdmdVWAgASABPS9z/TktQSwUGAAAAAAcA BwA2AgAAqwwAAAAA 55345 2010-05-07 00:13:40 -0700 2010-05-11 17:58:38 -0700 Patch (without optimization) svg-pattern-fill-bug.patch text/plain 1279 s3lance SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvQ2hhbmdlTG9n CShyZXZpc2lvbiA1ODk0MCkKKysrIFdlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBA IC0xLDMgKzEsMTIgQEAKKzIwMTAtMDUtMDcgIFN0ZXZlbiBMYWkgIDxzdGV2ZW5fbGFpQGFzaWEu YXBwbGUuY29tPgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAoT09QUyEpLgorCisgICAg ICAgIERvbid0IGNsaXAgdG8gdmlld3BvcnQgYW55bW9yZSBvbiBjcmVhdGluZyBhIHBhdHRlcm4g aW4gUmVuZGVyU1ZHUmVzb3VyY2VQYXR0ZXJuLmNwcAorCisgICAgICAgICogcmVuZGVyaW5nL1Jl bmRlclNWR1Jlc291cmNlUGF0dGVybi5jcHA6CisgICAgICAgIChXZWJDb3JlOjpSZW5kZXJTVkdS ZXNvdXJjZVBhdHRlcm46OmNyZWF0ZVRpbGVJbWFnZSk6CisKIDIwMTAtMDQtMjcgIFRvbnkgQ2hh bmcgIDx0b255QGNocm9taXVtLm9yZz4KIAogICAgICAgICBSZXZpZXdlZCBieSBFcmljIFNlaWRl bC4KSW5kZXg6IFdlYkNvcmUvcmVuZGVyaW5nL1JlbmRlclNWR1Jlc291cmNlUGF0dGVybi5jcHAK PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PQotLS0gV2ViQ29yZS9yZW5kZXJpbmcvUmVuZGVyU1ZHUmVzb3VyY2VQYXR0ZXJu LmNwcAkocmV2aXNpb24gNTg3ODkpCisrKyBXZWJDb3JlL3JlbmRlcmluZy9SZW5kZXJTVkdSZXNv dXJjZVBhdHRlcm4uY3BwCSh3b3JraW5nIGNvcHkpCkBAIC0yMzksOSArMjM5LDYgQEAgUGFzc093 blB0cjxJbWFnZUJ1ZmZlcj4gUmVuZGVyU1ZHUmVzb3VyYwogCiAgICAgSW50U2l6ZSBpbWFnZVNp emUobHJvdW5kZihwYXR0ZXJuQm91bmRhcmllc0luY2x1ZGluZ092ZXJmbG93LndpZHRoKCkpLCBs cm91bmRmKHBhdHRlcm5Cb3VuZGFyaWVzSW5jbHVkaW5nT3ZlcmZsb3cuaGVpZ2h0KCkpKTsKIAot ICAgIC8vIEZJWE1FOiBXZSBzaG91bGQgYmUgYWJsZSB0byBjbGlwIHRoaXMgbW9yZSwgbmVlZHMg aW52ZXN0aWdhdGlvbgotICAgIGNsYW1wSW1hZ2VCdWZmZXJTaXplVG9WaWV3cG9ydChvYmplY3Qt PmRvY3VtZW50KCktPnZpZXcoKSwgaW1hZ2VTaXplKTsKLQogICAgIC8vIERvbid0IGNyZWF0ZSBJ bWFnZUJ1ZmZlcnMgd2l0aCBpbWFnZSBzaXplIG9mIDAKICAgICBpZiAoaW1hZ2VTaXplLmlzRW1w dHkoKSkKICAgICAgICAgcmV0dXJuIDA7Cg== 55758 2010-05-11 14:35:55 -0700 2010-06-21 19:30:14 -0700 Patch 38704.patch text/plain 4044 s3lance SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvQ2hhbmdlTG9n CShyZXZpc2lvbiA1OTE3MSkKKysrIFdlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBA IC0xLDMgKzEsMjIgQEAKKzIwMTAtMDUtMTEgIFN0ZXZlbiBMYWkgIDxzdGV2ZW5fbGFpQGFzaWEu YXBwbGUuY29tPgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAoT09QUyEpLgorCisgICAg ICAgIFNWRyBwYXR0ZXJuIHNpemUgY2hhbmdlZCB3aGVuIHJlc2l6aW5nIGJyb3dzZXIKKyAgICAg ICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTM4NzA0CisgICAgICAg IDxyZGFyOi8vcHJvYmxlbS83NTQxNDY2PgorCisgICAgICAgIENsYW1wIHBhdHRlcm4gdGlsZSBp bWFnZSBzaXplIGFnYWluc3QgdGhlIGRvY3VtZW50IGluc3RlYWQgb2YgY2xhbXBpbmcgYWdhaW5z dCB0aGUgdmlld3BvcnQgc2l6ZQorICAgICAgICBCdXQgd2Ugc2hvdWxkIGJlIGNsYW1waW5nIGFn YWluc3QgZWFjaCByZW5kZXIgb2JqZWN0IHRoYXQgdXNlcyB0aGlzIHBhdHRlcm4gZmlsbCBpbnN0 ZWFkCisKKyAgICAgICAgTm8gbmV3IHRlc3RzLiAoT09QUyEpCisKKyAgICAgICAgKiByZW5kZXJp bmcvUmVuZGVyU1ZHUmVzb3VyY2VQYXR0ZXJuLmNwcDoKKyAgICAgICAgKFdlYkNvcmU6OlJlbmRl clNWR1Jlc291cmNlUGF0dGVybjo6Y3JlYXRlVGlsZUltYWdlKToKKyAgICAgICAgKiByZW5kZXJp bmcvU1ZHUmVuZGVyU3VwcG9ydC5jcHA6CisgICAgICAgIChXZWJDb3JlOjpjbGFtcEltYWdlQnVm ZmVyU2l6ZVRvRG9jdW1lbnQpOgorICAgICAgICAqIHJlbmRlcmluZy9TVkdSZW5kZXJTdXBwb3J0 Lmg6CisKIDIwMTAtMDUtMTEgIEJyaWFuIFdlaW5zdGVpbiAgPGJ3ZWluc3RlaW5AYXBwbGUuY29t PgogCiAgICAgICAgIFJldmlld2VkIGJ5IFN0ZXZlIEZhbGtlbmJ1cmcuCkluZGV4OiBXZWJDb3Jl L3JlbmRlcmluZy9SZW5kZXJTVkdSZXNvdXJjZVBhdHRlcm4uY3BwCj09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdl YkNvcmUvcmVuZGVyaW5nL1JlbmRlclNWR1Jlc291cmNlUGF0dGVybi5jcHAJKHJldmlzaW9uIDU5 MDk0KQorKysgV2ViQ29yZS9yZW5kZXJpbmcvUmVuZGVyU1ZHUmVzb3VyY2VQYXR0ZXJuLmNwcAko d29ya2luZyBjb3B5KQpAQCAtMjM5LDggKzIzOSw4IEBAIFBhc3NPd25QdHI8SW1hZ2VCdWZmZXI+ IFJlbmRlclNWR1Jlc291cmMKIAogICAgIEludFNpemUgaW1hZ2VTaXplKGxyb3VuZGYocGF0dGVy bkJvdW5kYXJpZXNJbmNsdWRpbmdPdmVyZmxvdy53aWR0aCgpKSwgbHJvdW5kZihwYXR0ZXJuQm91 bmRhcmllc0luY2x1ZGluZ092ZXJmbG93LmhlaWdodCgpKSk7CiAKLSAgICAvLyBGSVhNRTogV2Ug c2hvdWxkIGJlIGFibGUgdG8gY2xpcCB0aGlzIG1vcmUsIG5lZWRzIGludmVzdGlnYXRpb24KLSAg ICBjbGFtcEltYWdlQnVmZmVyU2l6ZVRvVmlld3BvcnQob2JqZWN0LT5kb2N1bWVudCgpLT52aWV3 KCksIGltYWdlU2l6ZSk7CisgICAgLy8gRklYTUU6IFdlIHNob3VsZCBiZSBjbGFtcGluZyBhZ2Fp bnN0IHRoZSBzaXplIG9mIHRoZSBkaW1lbnNpb25zIG9mIGVhY2ggcmVuZGVyIG9iamVjdCB0aGF0 IHJlZmVyZW5jZSB0aGlzIHBhdHRlcm4gZmlsbAorICAgIGNsYW1wSW1hZ2VCdWZmZXJTaXplVG9E b2N1bWVudChvYmplY3QtPmRvY3VtZW50KCksIGltYWdlU2l6ZSk7CiAKICAgICAvLyBEb24ndCBj cmVhdGUgSW1hZ2VCdWZmZXJzIHdpdGggaW1hZ2Ugc2l6ZSBvZiAwCiAgICAgaWYgKGltYWdlU2l6 ZS5pc0VtcHR5KCkpCkluZGV4OiBXZWJDb3JlL3JlbmRlcmluZy9TVkdSZW5kZXJTdXBwb3J0LmNw cAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09Ci0tLSBXZWJDb3JlL3JlbmRlcmluZy9TVkdSZW5kZXJTdXBwb3J0LmNwcAko cmV2aXNpb24gNTkwOTQpCisrKyBXZWJDb3JlL3JlbmRlcmluZy9TVkdSZW5kZXJTdXBwb3J0LmNw cAkod29ya2luZyBjb3B5KQpAQCAtMjE0LDE5ICsyMTQsMjMgQEAgdm9pZCByZW5kZXJTdWJ0cmVl VG9JbWFnZShJbWFnZUJ1ZmZlciogaQogICAgICAgICBzdmdDb250YWluZXItPnNldERyYXdzQ29u dGVudHMoZmFsc2UpOwogfQogCi12b2lkIGNsYW1wSW1hZ2VCdWZmZXJTaXplVG9WaWV3cG9ydChG cmFtZVZpZXcqIGZyYW1lVmlldywgSW50U2l6ZSYgc2l6ZSkKK3ZvaWQgY2xhbXBJbWFnZUJ1ZmZl clNpemVUb0RvY3VtZW50KERvY3VtZW50ICpkb2N1bWVudCwgSW50U2l6ZSYgc2l6ZSkKIHsKLSAg ICBpZiAoIWZyYW1lVmlldykKKyAgICBpZiAoIWRvY3VtZW50LT52aWV3KCkpCiAgICAgICAgIHJl dHVybjsKIAotICAgIGludCB2aWV3V2lkdGggPSBmcmFtZVZpZXctPnZpc2libGVXaWR0aCgpOwot ICAgIGludCB2aWV3SGVpZ2h0ID0gZnJhbWVWaWV3LT52aXNpYmxlSGVpZ2h0KCk7CisgICAgY29u c3QgSW50U2l6ZSYgY29udGVudHNTaXplID0gZG9jdW1lbnQtPnZpZXcoKS0+Y29udGVudHNTaXpl KCk7CiAKLSAgICBpZiAoc2l6ZS53aWR0aCgpID4gdmlld1dpZHRoKQotICAgICAgICBzaXplLnNl dFdpZHRoKHZpZXdXaWR0aCk7CisgICAgaW50IHNjcm9sbFdpZHRoID0gZG9jdW1lbnQtPmRvY3Vt ZW50RWxlbWVudCgpLT5zY3JvbGxXaWR0aCgpOworICAgIGludCBzY3JvbGxIZWlnaHQgPSBkb2N1 bWVudC0+ZG9jdW1lbnRFbGVtZW50KCktPnNjcm9sbEhlaWdodCgpOworICAgIGludCB3ID0gY29u dGVudHNTaXplLndpZHRoKCkgPiBzY3JvbGxXaWR0aCA/IGNvbnRlbnRzU2l6ZS53aWR0aCgpIDog c2Nyb2xsV2lkdGg7CisgICAgaW50IGggPSBjb250ZW50c1NpemUuaGVpZ2h0KCkgPiBzY3JvbGxI ZWlnaHQgPyBjb250ZW50c1NpemUuaGVpZ2h0KCkgOiBzY3JvbGxIZWlnaHQ7CiAKLSAgICBpZiAo c2l6ZS5oZWlnaHQoKSA+IHZpZXdIZWlnaHQpCi0gICAgICAgIHNpemUuc2V0SGVpZ2h0KHZpZXdI ZWlnaHQpOworICAgIGlmIChzaXplLndpZHRoKCkgPiB3KQorICAgICAgICBzaXplLnNldFdpZHRo KHcpOworCisgICAgaWYgKHNpemUuaGVpZ2h0KCkgPiBoKQorICAgICAgICBzaXplLnNldEhlaWdo dChoKTsKIH0KIAogRmxvYXRSZWN0IFNWR1JlbmRlckJhc2U6OmNvbXB1dGVDb250YWluZXJCb3Vu ZGluZ0JveChjb25zdCBSZW5kZXJPYmplY3QqIGNvbnRhaW5lciwgYm9vbCBpbmNsdWRlQWxsUGFp bnRlZENvbnRlbnQpCkluZGV4OiBXZWJDb3JlL3JlbmRlcmluZy9TVkdSZW5kZXJTdXBwb3J0LmgK PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PQotLS0gV2ViQ29yZS9yZW5kZXJpbmcvU1ZHUmVuZGVyU3VwcG9ydC5oCShyZXZp c2lvbiA1OTA5NCkKKysrIFdlYkNvcmUvcmVuZGVyaW5nL1NWR1JlbmRlclN1cHBvcnQuaAkod29y a2luZyBjb3B5KQpAQCAtODIsNyArODIsNyBAQCB2b2lkIGFwcGx5VHJhbnNmb3JtVG9QYWludElu Zm8oUmVuZGVyT2JqCiB2b2lkIHJlbmRlclN1YnRyZWVUb0ltYWdlKEltYWdlQnVmZmVyKiwgUmVu ZGVyT2JqZWN0Kik7CiAKIHZvaWQgZGVyZWdpc3RlckZyb21SZXNvdXJjZXMoUmVuZGVyT2JqZWN0 Kik7Ci12b2lkIGNsYW1wSW1hZ2VCdWZmZXJTaXplVG9WaWV3cG9ydChGcmFtZVZpZXcqLCBJbnRT aXplJiBpbWFnZUJ1ZmZlclNpemUpOwordm9pZCBjbGFtcEltYWdlQnVmZmVyU2l6ZVRvRG9jdW1l bnQoRG9jdW1lbnQqLCBJbnRTaXplJiBpbWFnZUJ1ZmZlclNpemUpOwogCiB2b2lkIGFwcGx5U3Ry b2tlU3R5bGVUb0NvbnRleHQoR3JhcGhpY3NDb250ZXh0KiwgY29uc3QgUmVuZGVyU3R5bGUqLCBj b25zdCBSZW5kZXJPYmplY3QqKTsKIERhc2hBcnJheSBkYXNoQXJyYXlGcm9tUmVuZGVyaW5nU3R5 bGUoY29uc3QgUmVuZGVyU3R5bGUqIHN0eWxlLCBSZW5kZXJTdHlsZSogcm9vdFN0eWxlKTsK 55760 2010-05-11 14:48:50 -0700 2010-05-11 14:48:50 -0700 Test case pattern-internal-clip.svg image/svg+xml 2061 krit PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHhtbG5zOnhsaW5rPSJodHRw Oi8vd3d3LnczLm9yZy8xOTk5L3hsaW5rIj4KPGRlZnM+CjwhLS0gbm8gY3Jhc2ggaWYgd2l0aC9o ZWlnaHQgaXMgMCAtLT4KPHBhdHRlcm4gaWQ9InBhdHRlcm4tMSIgd2lkdGg9IjAiIGhlaWdodD0i MTAwIj4KICAgIDxyZWN0IHdpZHRoPSIxMDAwMCIgaGVpZ2h0PSIxMDAwMCIgZmlsbD0icmVkIi8+ CjwvcGF0dGVybj4KPCEtLSBjbGlwIHRvIHRhcmdldHMgc3Ryb2tlIGJvdW5kYXJpZXMsIG5vIHRy YW5zZm9ybWF0aW9uIC0tPgo8cGF0dGVybiBpZD0icGF0dGVybi0yIiB3aWR0aD0iMTAwMDAiIGhl aWdodD0iMTAwMDAiPgogICAgPHJlY3Qgd2lkdGg9IjEwMDAwIiBoZWlnaHQ9IjEwMDAwIiBmaWxs PSJncmVlbiIvPgo8L3BhdHRlcm4+CjwhLS0gY2xpcCB0byB0YXJnZXRzIHN0cm9rZSBib3VuZGFy aWVzLCB3aXRoIHNjYWxlIC0tPgo8cGF0dGVybiBpZD0icGF0dGVybi0yIiB3aWR0aD0iMTAwMDAi IGhlaWdodD0iMTAwMDAiIHBhdHRlcm5UcmFuc2Zvcm09InNjYWxlKDAuMDAxKSI+CiAgICA8cmVj dCB3aWR0aD0iMTAwMDAiIGhlaWdodD0iMTAwMDAiIGZpbGw9ImdyZWVuIi8+CjwvcGF0dGVybj4K PCEtLSBjbGlwIHRvIHRhcmdldHMgc3Ryb2tlIGJvdW5kYXJpZXMsIHdpdGggc2tldyg0NSkgLS0+ CjxwYXR0ZXJuIGlkPSJwYXR0ZXJuLTMiIHdpZHRoPSIxMDAiIGhlaWdodD0iMTAwIiBwYXR0ZXJu VHJhbnNmb3JtPSJza2V3WCg0NSkiPgogICAgPHJlY3Qgd2lkdGg9IjEwMCIgaGVpZ2h0PSIxMDAi IGZpbGw9ImdyZWVuIi8+CjwvcGF0dGVybj4KPCEtLSBjbGlwIHRvIHRhcmdldHMgc3Ryb2tlIGJv dW5kYXJpZXMsIHdpdGggdHJhbnNsYXRlIGFuZCBza2V3IC0tPgo8cGF0dGVybiBpZD0icGF0dGVy bi00IiB3aWR0aD0iMTAwIiBoZWlnaHQ9IjEwMCIgcGF0dGVyblRyYW5zZm9ybT0idHJhbnNsYXRl KC01MCwwKSBza2V3WCg0NSkiPgogICAgPHJlY3Qgd2lkdGg9IjEwMCIgaGVpZ2h0PSIxMDAiIGZp bGw9ImdyZWVuIi8+CjwvcGF0dGVybj4KPCEtLSBjbGlwIHRvIHRhcmdldHMgc3Ryb2tlIGJvdW5k YXJpZXMsIHdpdGggdHJhbnNsYXRlIGFuZCBza2V3IC0tPgo8cGF0dGVybiBpZD0icGF0dGVybi01 IiB3aWR0aD0iMTAwIiBoZWlnaHQ9IjEwMCIgcGF0dGVyblRyYW5zZm9ybT0icm90YXRlKDQ1LCA1 MCwgNTApIj4KICAgIDxyZWN0IHdpZHRoPSIxMDAiIGhlaWdodD0iMTAwIiBmaWxsPSJncmVlbiIv Pgo8L3BhdHRlcm4+CjwhLS0gY2xpcCB0byB0YXJnZXRzIHN0cm9rZSBib3VuZGFyaWVzLCB3aXRo IHNjYWxlLCByb3RhdGUgLS0+CjxwYXR0ZXJuIGlkPSJwYXR0ZXJuLTYiIHdpZHRoPSIxMDAwMCIg aGVpZ2h0PSIxMDAwMCIgcGF0dGVyblRyYW5zZm9ybT0ic2NhbGUoMC4wMSkgdHJhbnNsYXRlKDUw MDAsIC0yNTAwKSByb3RhdGUoNDUsIDUwLCA1MCkiPgogICAgPHJlY3Qgd2lkdGg9IjEwMDAwIiBo ZWlnaHQ9IjEwMDAwIiBmaWxsPSJncmVlbiIvPgo8L3BhdHRlcm4+CjwvZGVmcz4KCjxyZWN0IHdp ZHRoPSIxMDAiIGhlaWdodD0iMTAwIiBmaWxsPSJ1cmwoI3BhdHRlcm4tMSkgZ3JlZW4iIHRyYW5z Zm9ybT0idHJhbnNsYXRlKDEwLDEwKSIvPgo8cmVjdCB3aWR0aD0iMTAwIiBoZWlnaHQ9IjEwMCIg ZmlsbD0idXJsKCNwYXR0ZXJuLTIpIiB0cmFuc2Zvcm09InRyYW5zbGF0ZSgxMzAsMTApIi8+Cjxy ZWN0IHdpZHRoPSIxMDAiIGhlaWdodD0iMTAwIiBmaWxsPSJ1cmwoI3BhdHRlcm4tMykiIHRyYW5z Zm9ybT0idHJhbnNsYXRlKDEwLDEzMCkiLz4KPHJlY3Qgd2lkdGg9IjEwMCIgaGVpZ2h0PSIxMDAi IGZpbGw9InVybCgjcGF0dGVybi00KSIgdHJhbnNmb3JtPSJ0cmFuc2xhdGUoMTMwLDEzMCkiLz4K PHJlY3Qgd2lkdGg9IjEwMCIgaGVpZ2h0PSIxMDAiIGZpbGw9InVybCgjcGF0dGVybi01KSIgdHJh bnNmb3JtPSJ0cmFuc2xhdGUoMTAsMjUwKSIvPgo8cmVjdCB3aWR0aD0iMTAwIiBoZWlnaHQ9IjEw MCIgZmlsbD0idXJsKCNwYXR0ZXJuLTYpIiB0cmFuc2Zvcm09InRyYW5zbGF0ZSgxMzAsMjUwKSIv Pgo8L3N2Zz4K