December 11, 2013 
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >
December 11, 2013 NOTICE: This channel is no longer actively logged.
[00:29:04] *** bsd has joined #eclipse-e4[00:30:41] *** bsd has quit IRC[00:30:50] *** bsd has joined #eclipse-e4[00:34:59] *** bsd has quit IRC[00:35:06] *** bsd has joined #eclipse-e4[00:36:11] *** bsd1 has joined #eclipse-e4[00:36:11] *** bsd has quit IRC[00:37:55] *** bsd1 has quit IRC[00:37:55] *** bsd has joined #eclipse-e4[00:42:25] *** bsd has quit IRC[01:38:44] *** bsd has joined #eclipse-e4[01:43:26] *** bsd has quit IRC[02:39:28] *** bsd has joined #eclipse-e4[02:43:45] *** bsd has quit IRC[03:40:16] *** bsd has joined #eclipse-e4[03:44:33] *** bsd has quit IRC[04:41:01] *** bsd has joined #eclipse-e4[04:45:10] *** bsd has quit IRC[05:41:49] *** bsd has joined #eclipse-e4[05:46:04] *** bsd has quit IRC[06:42:36] *** bsd has joined #eclipse-e4[06:47:18] *** bsd has quit IRC[07:43:15] *** bsd has joined #eclipse-e4[07:47:39] *** bsd has quit IRC[08:13:18] *** drolka has joined #eclipse-e4[08:29:42] *** hendrikstill has quit IRC[08:38:23] *** hendrikstill has joined #eclipse-e4[08:40:34] *** meszarosi has joined #eclipse-e4[08:44:18] *** bsd has joined #eclipse-e4[08:48:46] *** bsd has quit IRC[08:59:08] *** bsd has joined #eclipse-e4[09:00:48] *** bsd has quit IRC[09:00:55] *** bsd has joined #eclipse-e4[09:05:43] *** bsd has quit IRC[09:10:58] *** m1hael has joined #eclipse-e4[09:20:28] *** tomschindl has joined #eclipse-e4[09:47:34] *** drolka1 has joined #eclipse-e4[09:49:45] *** drolka has quit IRC[10:01:40] *** bsd has joined #eclipse-e4[10:06:23] *** bsd has quit IRC[10:25:53] *** bsd has joined #eclipse-e4[10:30:27] *** wsudol has joined #eclipse-e4[12:15:16] *** hendrikstill has quit IRC[12:18:33] *** bsd has quit IRC[13:00:49] *** paulweb515 has quit IRC[13:06:40] *** hio has joined #eclipse-e4[14:19:18] *** emoffatt has joined #eclipse-e4[14:24:34] *** hendrikstill has joined #eclipse-e4[14:43:22] *** pelder has joined #eclipse-e4[14:44:02] *** hendrikstill has quit IRC[14:45:17] *** hendrikstill has joined #eclipse-e4[14:53:11] *** paulweb515 has joined #eclipse-e4[14:53:25] *** paulweb515 has quit IRC[14:53:40] *** paulweb515 has joined #eclipse-e4[15:20:14] <paulweb515> drolka1: ~423768[15:20:15] <Arbalest> Bug 423768 - http://bugs.eclipse.org/423768 - Platform / UI / 4.4 - PC / Windows 7 - ASSIGNED / / major / - Assignee: daniel.rolka - [CSS] IAE exception when switching between CSS stylesheets with combo[15:20:23] <paulweb515> drolka1: is this something that we need to fix today for M4?[15:20:36] <paulweb515> drolka1: we're allowed to fix serious bugs that we find when testing the M4 functionality[15:26:32] <drolka1> paulweb515: I don't think so since the CSS bridge is not enabled yet. We need the CSS stylesheet modification to enable it[15:27:14] <drolka1> paulweb515: and there is workaround for it that I've described in the bug[15:28:36] <paulweb515> drolka1: I know we didn't ship it, but we were going to tell other people in the community about it[15:28:59] <paulweb515> drolka1: is this going to then cause them trouble?[15:29:44] <drolka1> paulweb515: so in such case maybe would be better to push the fix for 423768[15:30:05] <paulweb515> OK, I'll try and review it today[15:30:12] <drolka1> paulweb515: OK, thanks[15:30:22] <paulweb515> well, if we want it in I'll have to review it today :-)[15:31:25] <paulweb515> drolka1: If we remove all of the resources from the cache, don't we just leak?[15:32:02] <drolka1> paulweb515: yes, it does, but the user has to restart the Eclipse to apply the new CSS theme[15:32:13] <drolka1> paulweb515: there is proper warning message for it[15:32:45] <paulweb515> drolka1: I have always been pushing managing the 3.x SWT resources in the CSS engine differently because I wanted to reuse the already safe 3.x fonts and colors. But since you've determined it's not appropriate to use the 3.x resources directly[15:33:04] <paulweb515> drolka1: then the CSS engine owns all of those resources as well, just like the other CSS resources[15:33:46] <paulweb515> drolka1: what if we removed the "removes and disposes" steps instead. That would get rid of the IAE and the CSS engine would be responsible for the resources in the SWTResourceRegistry just like the rest[15:36:05] <drolka1> paulweb515: are you asking about ThemeEngine change or the WBWRenderer?[15:37:04] <paulweb515> drolka1: I'm saying we would remove both of them, since the CSS engine wholey owns the fonts retrieved from color and font definitions[15:42:30] <drolka1> paulweb515: I think we should follow other direction - see the bug proposal: ~423813[15:42:31] <Arbalest> Bug 423813 - http://bugs.eclipse.org/423813 - Platform / UI / 4.4 - PC / Windows 7 - NEW / / normal / - Assignee: daniel.rolka - [CSS] Improvement of the CSS theme switching and applying the new stylesheet in runtime - proposal[15:45:32] *** hio has quit IRC[15:47:59] *** hio has joined #eclipse-e4[15:52:47] <paulweb515> drolka1: but that's just it. We don't have to dispose unused resources (at least not yet). We just have to cache them the same way other stuff is cached[15:53:12] <paulweb515> drolka1: AFAIK if you change CSS style sheets, you don't free up the Fonts/Colors from the last style sheet, they're still cached[15:53:27] <drolka1> paulweb515: what about memory leaks after modifying 3x theme?[15:53:36] <paulweb515> drolka1: we would just make ourselves consistent with that.[15:54:11] <paulweb515> drolka1: after modifying a 3.x theme don't you get back new FontData for that symbolic name?[15:55:21] <drolka1> paulweb515: but in the cache we keep the Font, Color, Image, Cursor instance (not descriptors like FontData, ImageData, ...)[15:55:53] <drolka1> paulweb515: let me try describe how it works, give me a moment[15:56:32] <paulweb515> drolka1: right. so the problem is it is attached to the key, and we need to store the new Font object against the same key (which has the old Font object)[15:56:53] <drolka1> paulweb515: yes[15:58:50] <paulweb515> OK, I think we need to do some more thinking in your ~423813[15:58:51] <Arbalest> Bug 423813 - http://bugs.eclipse.org/423813 - Platform / UI / 4.4 - PC / Windows 7 - NEW / / normal / - Assignee: daniel.rolka - [CSS] Improvement of the CSS theme switching and applying the new stylesheet in runtime - proposal[15:59:20] <paulweb515> drolka1: what about collecting the resources and then disposing of them after the engine.reset() ? is that not a similar pattern to what was done in WBWRenderer?[15:59:56] <drolka1> paulweb515: I was trying it, but it doesn't work since Widget still use resources[16:00:02] <paulweb515> drolka1: or the problem is because of the way CSS works, a widget that was styled in the old sheet but is not styled in the new sheet we have not way to "put it back to before it was styled"[16:02:06] <drolka1> paulweb515: we have to use some default resources, because now we use the resource from Widgets. When you style let say button, first you take a FontData from Button.getFont().getFontData()[] and next add style modification (size, style, ...)[16:02:51] <drolka1> paulweb515: so when resource that currently is kept by Widget is disposed and you call Button.getFont().getFontData() you get SWT exception[16:03:30] <paulweb515> The SWTResourceRegistry was designed to create a Font or Color and hold onto it for the lifetime of the Engine[16:04:31] <drolka1> paulweb515: yes, but now we apply changes in the runtime and update the cache[16:04:38] <paulweb515> So we're trying to swap out a Font or Color by ID when there might by widgets still using the old font[16:04:57] <drolka1> paulweb515: yes[16:05:46] <drolka1> paulweb515: for instance the new CSS you have selected doesn't have some style used by the old one (color for instance). So the new style uses old styles and the new ones[16:06:12] <drolka1> paulweb515: it is the reason you have to restart the Eclipse to fix the new style[16:06:16] <paulweb515> And because there's no "reset this widget to before it had CSS properties" so we can't reliably release any resource that we've used[16:06:58] *** wsudol has quit IRC[16:07:30] *** bsd has joined #eclipse-e4[16:10:46] <drolka1> paulweb515: for instance we have the issue here - CSSValueSWTFontDataConverterImpl.convert -> we get FontData of Font that can be disposed and we get SWT exception. It would be better to take some global default in such case. We need this data here: CSSSWTFontHelper.getFontData(CSS2FontProperties fontProperties, FontData oldFontData) -> oldFontData should be replaced with some global default to avoid SWT exception[16:12:49] <paulweb515> drolka1: when you get the IAE, what happens? Is the system unusable, or does it fix itself?[16:13:13] <drolka1> paulweb515: it is unsable[16:13:18] <paulweb515> drolka1: or is the only solution once you start getting the IAE that you restart?[16:13:34] <paulweb515> Can you re-apply the old CSS style sheet and fix it, or will the engine just yack[16:16:08] <drolka1> paulweb515: but are you asking me about runtime or what the user has to do?[16:16:55] <paulweb515> What the user has to do. Can the user go back to the dialog and re-apply the previous style sheet (that would have rules for the CTabItem in question) and does that make it go away[16:18:11] <drolka1> paulweb515: unfortunately they have to restart the entire Eclipse. Workbench is corrupted and unusable[16:19:50] <paulweb515> drolka1: OK[16:20:32] <paulweb515> drolka1: I guess we go ahead with 423768 as a temporary fix and then we will need to re-examine resource management in the engine early in M5[16:20:51] <drolka1> paulweb515: OK[16:21:12] <paulweb515> drolka1: JFace has a couple of classes that aid in resource management (this problem is not new :-)[16:22:06] <paulweb515> drolka1: JFaceResources is the central class that contains common registries. LocalResourceManagers can be used to create and destroy images, fonts, and colors keyd by *Descriptors ... it manages them behind the scene using ref counting[16:22:28] <drolka1> paulweb515: so maybe we can process the ~423813 to improve the CSS engine as well as the Theme switching to avoid similar issue in the future[16:22:29] <Arbalest> Bug 423813 - http://bugs.eclipse.org/423813 - Platform / UI / 4.4 - PC / Windows 7 - NEW / / normal / - Assignee: daniel.rolka - [CSS] Improvement of the CSS theme switching and applying the new stylesheet in runtime - proposal[16:22:46] <paulweb515> drolka1: I agree[16:23:05] <drolka1> paulweb515: I'm not sure if we can use the jFaceResource directly:[16:23:26] <drolka1> paulweb515: please take a look:[16:23:36] <drolka1> paulweb515: for such styling:[16:23:53] <drolka1> .MPart.busy {font-style: italic;}[16:23:53] <drolka1> .MPart.highlighted {font-weight: bold;}[16:23:53] <drolka1> .MPart {font-family: '#org-eclipse-ui-workbench-TAB_TEXT_FONT';}[16:24:00] <paulweb515> drolka1: we won't have a visibility of bundle problem, it's valid for css.swt to use org.eclipse.jface.[16:24:12] <drolka1> paulweb515: you have such cache:[16:24:25] <paulweb515> drolka1: we might have to enhance the key to be symbolicName-FontDescriptor[16:25:12] <drolka1> '#org-eclipse-ui-workbench-TAB_TEXT_FONT_null_null' -> Font instance 1[16:25:13] <drolka1> '#org-eclipse-ui-workbench-TAB_TEXT_FONT_null_bold' -> Font instance 2[16:25:13] <drolka1> '#org-eclipse-ui-workbench-TAB_TEXT_FONT_null_italic' -> Font instance 3[16:25:15] *** ircbog has joined #eclipse-e4[16:25:28] <paulweb515> drolka1: or manage another cache of fontkey to fontdescriptor[16:26:04] <paulweb515> FontDescriptors compare their FontData to manage equiality[16:26:38] <drolka1> for the MPart will be styled with 'Font instance1', when it contains 'busy' tag it will be the 'Font instance 2' and for 'highlighted' it will be 'Font instance 3'[16:28:08] <drolka1> paulweb515: drolka1: 'or manage another cache of fontkey to fontdescriptor' -> it sounds good to me, but it requires several modifications in the CSS engine[16:28:37] <paulweb515> drolka1: yes, that's why we should go ahead with the IAE fix now and consider you other bug in M5[16:28:51] <drolka1> paulweb515: OK[16:44:48] <paulweb515> drolka1: What else is needed over the last Classic theme that you pushed to Gerrit[16:44:58] <paulweb515> ?[16:45:41] <drolka1> paulweb515: ~423607[16:45:42] <Arbalest> Bug 423607 - http://bugs.eclipse.org/423607 - Platform / UI / 4.4 - PC / Windows 7 - ASSIGNED / / normal / - Assignee: daniel.rolka - [CSS] Gradient in background-color is not applied to children[16:46:30] <paulweb515> drolka1: I see that. I have the updated Classic CSS sheet in my inner. I'm asking what part of the style sheet in the bug will trigger the problem?[16:47:36] <drolka1> paulweb515: just put the following styles to the classic stylesheet:[16:47:36] <drolka1> .MPart.busy { font-style: italic; } .MPart.highlighted { font-weight: bold; } .MPartStack, .MPart { font-family: '#org-eclipse-ui-workbench-TAB_TEXT_FONT'; } CTabItem:selected { color: '#org-eclipse-ui-workbench-ACTIVE_TAB_TEXT_COLOR'; }[16:48:24] <paulweb515> drolka1: that's already in the version I have[16:48:42] <drolka1> paulweb515: and do you see the issue?[16:48:45] *** meszarosi has quit IRC[16:49:53] *** paulweb5151 has joined #eclipse-e4[16:49:56] <drolka1> paulweb515: you have to switch between classic theme and other several times. You can also try to update the 3.x theme in the meanwhile[16:49:57] *** paulweb5151 has quit IRC[16:50:03] *** paulweb515 has quit IRC[16:50:12] *** paulweb515 has joined #eclipse-e4[16:50:55] <paulweb515> No, I've tried switching between Classic and GTK a couple of times, and I don't see the issue[16:52:36] <drolka1> paulweb515: try the following path: switch to the classic then (with CSS bridge), next restart the Eclipse and modify 'Part title font' and start to switch the CSS themes[16:53:52] <drolka1> paulweb515: of course you have to apply the changes (apply or OK button) after switching to the new CSS theme[16:54:17] <paulweb515> yes, I still don't see the changes[16:54:52] <paulweb515> drolka1: OK, I didn't change the part title font, just a sec[16:56:03] *** ircbog has left #eclipse-e4[16:56:48] <paulweb515> drolka1: OK, that did it. Thanks[16:58:18] <drolka1> paulweb515: np, have you tried my fix for that?[16:59:20] *** hio has quit IRC[16:59:30] <paulweb515> I'm just testing it :-) I needed to be able to reproduce the problem first[17:14:33] *** pelder has left #eclipse-e4[17:24:42] *** tomschindl has quit IRC[17:32:11] *** pelder has joined #eclipse-e4[17:36:14] *** emoffatt has quit IRC[18:02:39] *** user258467 has quit IRC[18:03:21] *** drolka1 has left #eclipse-e4[19:06:59] *** bsd has quit IRC[19:45:35] *** bsd has joined #eclipse-e4[19:45:47] *** bsd has quit IRC[20:35:17] *** pelder has quit IRC[20:46:12] *** bsd has joined #eclipse-e4[20:50:38] *** bsd has quit IRC[21:06:15] *** paulweb515 has quit IRC[21:33:43] *** hendrikstill has quit IRC[21:34:17] *** hendrikstill has joined #eclipse-e4[21:41:15] *** paulweb515 has joined #eclipse-e4[21:47:05] *** bsd has joined #eclipse-e4[21:48:54] *** bsd1 has joined #eclipse-e4[21:48:55] *** bsd has quit IRC[21:53:27] *** bsd1 has quit IRC[22:49:39] *** bsd has joined #eclipse-e4[22:54:41] *** bsd has quit IRC[23:50:25] *** bsd has joined #eclipse-e4[23:54:44] *** bsd has quit IRC[23:57:49] *** ircbog1 has joined #eclipse-e4[23:57:52] *** ircbog1 has left #eclipse-e4