Skip to main content
AgeCommit message (Collapse)AuthorFilesLines
2019-04-07Bug 297510 - Wait for the shell to become active in focus test casesRolf Theunissen4-22/+17
Fix failing tests in Cocoa, on GTK it replaces forceActivate of the shell Change-Id: Ifd27c4885f65a8aa2576977283a9b63a10d7abc2 Signed-off-by: Rolf Theunissen <>
2019-04-05Bug 545073 - [GTK] MouseNavigation pastes text on middle clickLaurent CARON2-16/+38
Change-Id: I7fa01f6520c43acd5dd39e5ba1ead6d5e2de9b0e Signed-off-by: Laurent CARON <>
2019-04-05Bug 297510 - Rework focus tests to make them pass on all platformsRolf Theunissen5-7/+54
- Use Text instead of Button for focus child - Drain event queue in tearDown on Cocoa to close shells properly - Force focus on shell only on GTK - Disable focus test for some custom widgets Change-Id: I876f625fa5d00e8c1c0485efb675743540f02bee Signed-off-by: Rolf Theunissen <>
2019-04-04Bug 182598 - adjusted patch to deal with nested tree elementsSimeon Andreev2-29/+38
The previous patch for bug 182598 did not take tree items with parent tree items into account. This led to throwing exceptions in cases which were not broken. With this change, the tree elements that are being removed are correctly iterated. Change-Id: I8632202d13995f0cc2203a52027b0deb7006085a Signed-off-by: Simeon Andreev <> Signed-off-by: Andrey Loskutov <>
2019-04-03Bug 546096 - [GTK3] TextEditor toolbar icon wrong transparencyXi Yan1-4/+22
Add additional means of getting transparencyMask from alphaData. We should not get mask from transparentPixel if we already have alphaData. Change-Id: I546a4a120babbfe28e7e7d9dfee2c2b75569d3dc Signed-off-by: Xi Yan <>
2019-04-03Bug 546089 - [GTK3] Jagged edges in some iconsXi Yan1-2/+0
Revert Better solution found in Change-Id: I1ea0d07d9450dc52b09f87784d8b673779831d53 Signed-off-by: Xi Yan <>
2019-04-03Bug 546035 - [GTK3][regression] Some icons aren't rendered properlyXi Yan1-22/+61
anymore Caused by ImageLoader#save not working properly when we want to save in RGBA channel but ImageData is in RGB channel. Fix is to iterate over each pixel and fix the byte ordering to get a good pixel data buffer to pass to gdk_pixbuf_new. Tested with child Eclipse on X11 and Wayland with hover icons. Tested with ImageAnalyzer: loading any format image and saving it as PNG should work. Load JPEG file (in RGB), add Alpha mask to it, then saving it as PNG should work. Change-Id: I60dd32e313276ebb0dc35a5646beb4b57691e665 Signed-off-by: Xi Yan <>
2019-04-03Bug 545942: [GTK4] Ensure snippets work on GTK4Eric Williams1-2/+9
Remove last reference to gdk_window_get_display() on GTK4 to make Snippet39 run without crashing. Change-Id: I27886412cb67e713c760e29ed903f79b6cdba3ab Signed-off-by: Eric Williams <>
2019-04-03Bug 546073 - [GTK3] Warnings building bindings with GdkPixbuf functionsXi Yan2-14/+14
Remove const qualifier Change-Id: Ifa758742a6bf036b5f71164d3a92a23ccdb3abe8 Signed-off-by: Xi Yan <>
2019-04-03Bug 182598 - assertion failed in gtktreestore.cSimeon Andreev5-2/+147
Throw an exception instead of crashing, whenever a tree or table item is destroyed while its properties are being applied. Change-Id: Ia333a69f7ed94b5eed547a9f90120135f92b6eaa Signed-off-by: Simeon Andreev <> Signed-off-by: Andrey Loskutov <>
2019-04-03Bug 543704: SWT Table does not release all relevant items when input hasLakshmi Shanmugam1-0/+55
changed Updated Table.remove tests to ensure that the removed items are disposed and other items are not. Change-Id: I226e7d49e2578fe5b6777c6fa5baf5be1d8dcf5b
2019-04-03Bug 536468: NPE when adding 2nd entry in module pathLakshmi Shanmugam1-5/+5
Table.remove doesn't set null to all the disposed items. Change-Id: I90ac6133b28c963102fcb0ec88e231b4d8ae86a3
2019-04-02Bug 545793 - [GTK] Remove 32-bit codeXi Yan28-436/+379
Remove more /*int*/ casts. Change-Id: I23076cf3cf9fc3fd87f59f7b36004016e41875ac Signed-off-by: Xi Yan <>
2019-04-02Fix Java 9 deprecation warningsNikita Nemkin12-28/+26
Boolean/Integer/Long/Double constructors were deprecated in Java 9. Change-Id: I30d054db50310de6b2ca4457d2548851b2e5e789 Signed-off-by: Nikita Nemkin <>
2019-04-02Bug 297510 - [Widgets] Calling Composite#setFocus() before shell isRolf Theunissen8-32/+57
opened sets focus to nowhere Instead of calling forceFocus, call setFocus on when the focus is restored, GTK already uses this behavior. This will propagate the focus to a child of a Container, when a shell becomes visible and active. Change-Id: I1dccdf08820342d6d6253d50ac76e5c41ff0d76b Signed-off-by: Rolf Theunissen <>
2019-04-01Bug 545994 - [GTK3][Tree] Visible area moved when only expanding elementXi Yan3-11/+12
Fix tree jump regression from bug 489751. Do not set Tree selection to first item. Keep track of TreeItem parent and deselect GtkTreeSelection after item is disposed. Remove /*int*/ cast. Tested with Bug489751_TreeItemDisposeSelect, Eclipse on X11 and Wayland. Change-Id: Ice0fbe7d4e656ce63a89cddf2fba5c990c26e085 Signed-off-by: Xi Yan <>
2019-03-29Bug 345972 - Transparent pixels lost when copying image that wasXi Yan2-103/+63
directly read from file Bug 103691 - Misidentification of transparent pixels Fix issue with transparent pixels lost by explicitly setting ImageData.transparentPixel=0 for images with having transparent pixels (i.e. alpha=0) in getImageData. Tested with TestBed in Bug 34592 and Bug103691_TransparentPixel. Change-Id: I9c286dbab37e8b362a7832e7ff3276cacb682d16 Signed-off-by: Xi Yan <>
2019-03-29Bug 545032 - [GTK] Implement native ImageLoaderXi Yan10-4/+1418
Implement ImageLoader on GTK with native GdkPixbuf API rather than using Java implementation. Loading images: 1) Load InputStream (or FileInputStream) into byte array 2) Write byte array to GdkPixbufLoader 3) Get GdkPixbufAnimation from loader 4) Use pixbufToImageData method to convert GdkPixbuf(s) from GdkPixbufAnimation into ImageData objects to store. Known issue: * GdkPixbuf abstracts away differences between interlaced v.s. regular images, which means ImageLoaderEvent cannot be triggered for partially loaded images. This breaks test_addImageLoaderListenerLorg_eclipse_swt_graphics_ImageLoaderListener:d. The workaround is to explicitly check the 29th byte of a PNG file, if it is zero then it is not interlaced so no ImageLoaderEvent is triggered. Saving images 1) implemented conversion from ImageData to GdkPixbuf 2) use gdk_pixbuf_save_to_bufferv to write to pixbuf to byte array which is used to write to OutputStream * Currently supports all image formats specified in SWT API, could be extended to more formats that GdkPixbuf API supports. Performance: - Loading 2MB image takes ~100ms with native implementation where as before takes ~300ms with java implementation Change-Id: If8fdc3e06f36d8747a7ccdfa1940751a9d196757 Signed-off-by: Xi Yan <>
2019-03-28Bug 541376 - [GTK4] Replace signals which have been removedXi Yan2-1/+7
Move 'change-value' signal from GtkScrollBar to be handled by 'scroll-child' in GtkScrollWindow. Change-Id: I66663f26ed718e492114a00bc048ea288ee722cc Signed-off-by: Xi Yan <>
2019-03-27Bug 545793: [GTK] Remove 32-bit codeEric Williams2-74/+2
Remove 32 bit from build scripts. Change-Id: Ib9c66468ae30a0a25045c95f7e606b8d404c09fd Signed-off-by: Eric Williams <>
2019-03-26Bug 54579: [GTK] Remove 32-bit codeEric Williams34-223/+223
Remove 64 int casts. Change-Id: Ie6ccaa9a2d056a934cc9813415fa86b12cf68b5a Signed-off-by: Eric Williams <>
2019-03-26Bug 545587 - [GTK] Incorrect COLOR_INFO_FOREGROUND on Ubuntu 18.10Alexandr Miloslavskiy5-39/+293
Fix 1: ------ Old approach no longer works, because GTK changed the way of specifying CSS classes. I think it happened in this GTK commit: 6f6e5365 by Emmanuele Bassi at 2015-11-08 18:10:00 Use CSS node name for GtkTooltipWindow instead of a style class. The new way is to use 'GtkWidgetPath' and set CSS name explicitly, see 'gtk_widget_class_set_css_name' in 'gtk_tooltip_window_class_init'. Fix 2: ------ 'styleContextGetBackgroundColor' now uses ARGB surface to obtain correct alpha component of the color. Fix 3: ------ 'styleContextGetBackgroundColor' now correctly works around premultiplied colors in cairo. Refactoring: ------------ 1) Renamed 'getBackgroundColor' 'styleContextGetBackgroundColor' to match the other function, 'styleContextGetColor' Change-Id: Ie93fa71addd16133989be81ebf8e0d0fb1e55e12 Signed-off-by: Alexandr Miloslavskiy <>
2019-03-26Bug 545793: [GTK] Remove 32-bit codeEric Williams153-13855/+10967
Remove /*int*/ casts. Change-Id: I9e8da81d6435312acdd06cd1b3e471275ca842a8 Signed-off-by: Eric Williams <>
2019-03-26Bug 488855 - [Win32] Browser widget under windows calls a method in the ↵Alexandr Miloslavskiy2-11/+21
wrong thread The problem is that Internet Explorer sometimes calls 'WebSite#MapUrlToZone' from non-main thread. This causes errors in 'Widget#checkWidget' when 'WebSite' tries to get to its 'IE' parent via .getParent(). In order to solve the problem I removed the code in question; now 'IE' tells 'WebSite' what to do. I also noticed that there are other similar pieces of code, for example in 'WebSite#ProcessUrlAction'. However, it uses '_getUrl().startsWith(IE.ABOUT_BLANK)' this time. I found that 'WebSite#MapUrlToZone' also used this approach previously, but it was changed due to performance reasons. Still, it's not clear whether all instances of '_getUrl().startsWith(IE.ABOUT_BLANK)' can be changed safely to 'IE#isAboutBlank' and I don't have the time to test it, so I left it as is. I tried to test the original problem for which 'WebSite#MapUrlToZone' was made, but it no longer reproduces for me. Probably it was solved independently in new IE. Change-Id: Icd27bb6481e4f6792fc844380d0da9dbdc82adff Signed-off-by: Alexandr Miloslavskiy <>
2019-03-25Bug 544498 - Call GTK.gtk_widget_get_window() directly to avoid ↵Marco Maccaferri1-1/+1
StackOverflowError Change-Id: Iaf4fd9283e67f36ea7ea15c3969f883a796fe2da Signed-off-by: Marco Maccaferri <>
2019-03-22Bug 545268 - Recreating TabItem crashes jvmPavel Matveev1-3/+7
Restore parent for control associated with tab item See Change-Id: I8c77475ab07ee42b6c1dc0db4b4347530a2c450c Signed-off-by: Pavel Matveev <>
2019-03-22Bug 545287 - Text is clippedPavel Matveev1-1/+1
Reduced minimum width for unbordered text. See Change-Id: I61829cf3e5a91608354a1d3479a66dae02a8f478 Signed-off-by: Pavel Matveev <>
2019-03-21Bug 545645 - [GTK] Table/Tree header must not keep focusXi Yan2-0/+2
Set table/tree header button focus on click to false. Change-Id: If618e1e1b0d47209fd1634f34c6693287a14a11d Signed-off-by: Xi Yan <>
2019-03-21Bug 545271 - Invisible Spinner text selectionPavel Matveev1-2/+14
Set spinner background See Change-Id: I1bbc56b14d4d51cd1ff9883b7d9fe704ae5d8df9 Signed-off-by: Pavel Matveev <>
2019-03-21Bug 545619 - [GTK] Building native libraries fails to auto-detect JAVA_HOME ↵Alexandr Miloslavskiy1-7/+7
for linux.x86_64 Change-Id: Ia5c30604366c790c5a679451c18a2845a9985fbb Signed-off-by: Alexandr Miloslavskiy <>
2019-03-21Bug 545610 - [GTK] Syntax error causing problems in JNIGeneratorAlexandr Miloslavskiy2-2/+2
Change-Id: I3347a2ff20ce13aa9d098abcea4ff48efe85996f Signed-off-by: Alexandr Miloslavskiy <>
2019-03-21Bug 545528 - JavaDoc for StyledTextContent#getLineAtOffset(int) isPaul Pazderski1-3/+25
contradictory JavaDoc defines getLineAtOffset(getLength()) should return number of lines. This contradicts the example in same JavaDoc which claims getLineAtOffset(4) == 2 (and the examples length is 4) but getLineCount returns 3. Changed JavaDoc to define getLineAtOffset(getLength()) as index of last line since this matches implementation of DefaultContent and most (or all?) other implementations. Change-Id: I9cdd96177ad8c47f30fe248ba1813c3be5a48417 Signed-off-by: Paul Pazderski <>

Back to the top