Hi Prasanta,
Ok, lets understand this in terms of the possible cases: When newWidth > 0 || newHeight > 0 - then we should update the state anyway. This covers 3 cases in which we need update the state. The only other case is when newWidth < 0 && newHeight < 0 - as in the dimensions are not specified in the html. In this case, we need to fallback to image dimensions, which again will warrant a state update. So, keeping this in view, I did not write another condition to check for the width /height change to update the state. Hope this clarifies the logic. Thanks, Krishna From: Prasanta Sadhukhan Sent: Wednesday, March 6, 2019 4:01 PM To: Krishna Addepalli <[email protected]> Cc: Sergey Bylokhov <[email protected]>; [email protected] Subject: Re: <Swing Dev> [13] RFR: JDK-8218674 - HTML Tooltip with "img=src" on component doesn't show Hi Krishna, But in other area of the code, like 778 if (newWidth > 0) { 779 newState |= WIDTH_FLAG; 780 } we do check for width/height > 0 before updating the state, so I would expect to maintain consistency, we should do the same for 797 newState |= (WIDTH_FLAG | HEIGHT_FLAG); Regards Prasanta On 06-Mar-19 3:25 PM, Krishna Addepalli wrote: Hi Prasanta, Handling 0-sized images was not done in the earlier fix as well. And this fix simply makes sure that the earlier fix works exactly the same when the image is loaded synchronously/asynchronously. At least from the looks of it, I do not see any harm in updating the image size, since it will only load the appropriate image, and update the ImageView object state. Thanks, Krishna On 05-Mar-2019, at 8:54 AM, Prasanta Sadhukhan <HYPERLINK "mailto:[email protected]"[email protected]> wrote: Hi Krishna, On 04-Mar-19 5:49 PM, Krishna Addepalli wrote: Hi Prasanta, Yes, you are right. But, does it make a difference if we still update the values even if they are same? I am not pointing to d.width = newWidth; My concern is newState |= (WIDTH_FLAG | HEIGHT_FLAG); where you update the state even if width/height can be 0. I am not sure about possible repurcussions? I do not see any test with image of width/height=0 so thought that you probably do not check that path. Regards Prasanta Thanks, Krishna On 04-Mar-2019, at 5:12 PM, Prasanta Sadhukhan <HYPERLINK "mailto:[email protected]"[email protected]> wrote: On 04-Mar-19 4:53 PM, Krishna Addepalli wrote: Hi Prasanta, Thanks for the review. Here is the updated webrev: HYPERLINK "http://cr.openjdk.java.net/%7Ekaddepalli/8218674/webrev01/"http://cr.openjdk.java.net/~kaddepalli/8218674/webrev01/ The check (newWidth>0) && (newHeight > 0) is done in adjustWidthHeight function. if (specifiedWidth != -1 && specifiedHeight != -1) { 915 newWidth = specifiedWidth; 916 newHeight = specifiedHeight; 917 } ... 931 d.width = newWidth; 932 d.height = newHeight; If specifiedWidth/Height is 0, then I do not see any check, it is just assigning to d.width/height, no? In that case, can we need to change the newstate without checking? Regards Prasanta Thanks, Krishna On 04-Mar-2019, at 1:29 PM, Prasanta Sadhukhan <HYPERLINK "mailto:[email protected]"[email protected]> wrote: Hi Krishna, You can reuse existing method getLoadsSynchronously() instead of checking for (state & SYNC_LOAD_FLAG) != 0) BTW, do we not have to check if (newWidth > 0) & newHeight >0 before changing the newstate @l797? The copyright year in testcase should be changed to 2019. Regards Prasanta On 01-Mar-19 12:10 PM, Krishna Addepalli wrote: Thanks for the review Sergey. Can I have one more review? Prasanta maybe? Thanks, Krishna On 01-Mar-2019, at 3:40 AM, Sergey Bylokhov HYPERLINK "mailto:[email protected]"<[email protected]> wrote: Looks fine. On 21/02/2019 08:44, Sergey Bylokhov wrote: On 20/02/2019 22:21, Krishna Addepalli wrote: Hi Sergey, I have fixed the issue. Could you check now? Yes, it works now, I look to the fix. Thanks, Krishna -----Original Message----- From: Sergey Bylokhov Sent: Thursday, February 21, 2019 3:54 AM To: Krishna Addepalli HYPERLINK "mailto:[email protected]"<[email protected]>; HYPERLINK "mailto:[email protected]"[email protected] Subject: Re: <Swing Dev> [13] RFR: JDK-8218674 - HTML Tooltip with "img=src" on component doesn't show Hi, Krishna. Some links have wrong file permissions, "403 - Forbidden": HYPERLINK "http://cr.openjdk.java.net/%7Ekaddepalli/8218674/webrev00/raw_files/new/test/jdk/javax/swing/text/html/8218674/circle.png"http://cr.openjdk.java.net/~kaddepalli/8218674/webrev00/raw_files/new/test/jdk/javax/swing/text/html/8218674/circle.png HYPERLINK "http://cr.openjdk.java.net/%7Ekaddepalli/8218674/webrev00/raw_files/new/test/jdk/javax/swing/text/html/8218674/TooltipImageTest.java"http://cr.openjdk.java.net/~kaddepalli/8218674/webrev00/raw_files/new/test/jdk/javax/swing/text/html/8218674/TooltipImageTest.java On 20/02/2019 03:57, Krishna Addepalli wrote: Hi All, Please review a fix for the bug JDK-8218674: https://bugs.openjdk.java.net/browse/JDK-8218674 Webrev: HYPERLINK "http://cr.openjdk.java.net/%7Ekaddepalli/8218674/webrev00/"http://cr.openjdk.java.net/~kaddepalli/8218674/webrev00/ This is a regression introduced due to fix for JDK-8208638. The default behaviour for ImageView is to load an image asynchronously. Hence, it uses the ImageHandler::imageUpdate to get the updates to the image being loaded. That will set the width and height of the image view. ImageView::updateImageSize does not alter the width and height in this case. When a JToolTip is created and html text set as tooltip, internally, the image is requested to be loaded synchronously, and in this case, ImageView::updateImageSize is the only way to calculate the image size. Since the width and height were not specified in the tooltip, the image was not being drawn. The fix is to check if the image is requested to be loaded synchronously, and if so, then do the same calculation as for the fix for JDK-8208638, which will provide valid image width and height, additionally also taking care of the scaling issues fixed for JDK-8208638. I have tested the fix on Windows, Linux(Ubuntu) and Mac, and found that it is working. I have also run all the jtreg tests under the test/jdk/javax/swing/text/html, and found no new failures. Thanks, Krishna -- Best regards, Sergey.
