l20n discussions and .platform

Hi all, as we're starting to discuss things on l20n, I'm having a question on the right group to do so. We did create a google group separate from the mozilla ecosystem, because technically, l20n isn't bound to gecko. The gecko impl of course is rather tied to gecko, haha, and thus I consider .platform to be a good place to discuss those issues. The API discussion that gandalf started is kind of spanning both worlds, and I'm wondering if those platform engineers interested in the more general api discussion would be equally fine having those in the google group or would

L20n and platform - early questions

We (l10n-drivers) are moving forward with our goal to introduce new localization platform for Gecko codenamed L20n. In a shortest possible words, L20n is supposed to replace .properties and .dtd files in Gecko and shift the paradigm of what localization process in order to give localizers ability to build flexible and rich phrases that match their language. (think: genders, declensions, plural forms). From the platform perspective L20n file format (compiled) is a simple JS file with simple entities: this.name = "value"; or complex ones: this.__defineGetter__("test", fun

OOPP and Responsiveness

OOPP is coming together nicely and as we move from finding basic problems like crashes and hangs, I'd like for us to take a look at little look at performance. We don't have a clear way to compare performance between Firefox 3.6 and the 1.9.3a3pre nightlies, for example, when using plugins, but if there's any perceived change in responsiveness I would like your help in calling it out. Just to remind you, OOPP works in Windows and Linux trunk nightlies. So far our testing has involved trying out sites with popular plugins to see if they crash and checking what the user interaction i

Debugging Minefield in CDT on Linux

I'm trying to debug Minefield in Eclipse 3.5.1 (from Universe) CDT (6.0.2 from eclipse.org) on 64-bit Ubuntu Karmic. First, I followed the instructions at http://curtisb.posterous.com/building-firefox-under-eclipsecdt-or-i-gotta except I didn't move the hg repo after cloning, because the new CDT was OK with creating a project with an existing directory. I get Eclipse to build and index the project. However, when launching the app (with MOZ_NO_REMOTE=1, NO_EM_RESTART=1 and LD_LIBRARY_PATH & LIBRARY_PATH from https://developer.mozilla.org/en/Eclipse), the app terminates after p

Architecture for viewing web archives (MHTML and MAFF)

Hello, I'm working on the Mozilla Archive Format extension for Firefox and SeaMonkey [1], which among other things allows the user to display content stored in web archives, which are single-file containers for all the resources required to render a web page. In MAF, the user experience of viewing a web archive is similar to that of other embedded content like PDF. For example, if I typed these addresses in the URL bar... http://www.example.com/printed-document-with-all-images.pdf http://www.example.com/web-page-with-all-images.mht ....I would see a complete page in the conte

Layers and scrolling

[Posting to dev.platform since this cuts across layout, gfx, and other stuff] Currently layout constructs a new layer tree for every paint. For various reasons, but mainly to better utilize D3D and OpenGL, we need to move to a model where we keep the same layer tree between paints and update it as necessary. Here are some design notes on how I think we should implement that, with particular attention to the implementation of scrolling. == Current Approach To Layer Construction == The current approach takes a display list and builds a list of layers that render that displ

getElementById not working ?

Trying to get getElementById to work with the embedded browser control in VB. I can navigate to a page fine, but getElementById just reports "object does not support this property or method" so do I presume they did not "finish it off" with all the functions that are found in the IE controls ? Cheers, Chris

Need help for JavaScript file editing/updating and debugging : Trying to fix bug 429827

I am trying to fix the problem mentioned in https://bugzilla.mozilla.org/show_bug.cgi?id=429827 But in order to fix this, it turns out I need to fix a javascript (.js) file. I have fixed a few C++ files before (see bug https://bugzilla.mozilla.org/show_bug.cgi?id=387502 ), so I am more or less sure that I have proper source tree and compilation tools. But JavaScript is a totally new beast to me. I have a couple of questions which I am sure that someone can answer. Question-1: Why are there multiple copies of "nsHelperAppDlg.js" under a few different directories in the source

marvendas@gmail.com Kit completo de Solenoides ( solenoid ) + chicote Para Cambio automatico 01M hidramatico Audi A3 Vw Golf gti turbo 31303

Contato: marvendas@gmail.com marvendas @ gmail.com marvendas no gmail.com Kit completo de solenoides para Volkswagem e Audi. O kit contem: 5 solenoides 2 Epc ( solenoides de pressao ) 1 Chicote Serve para qualquer modelo VW ou Audi fabricados de 1995 ate hoje com o cambio automatico de 4 marchas � 01M Pre�o: R$ 1900.00 Temos outras tipos de solenoides e artigos importados, nao deixe de fazer uma consulta antes de comprar! Audi a3 automatico Audi a3 1.8 t automatico Audi a3 1.8 turbo automatico VW Golf gti automatico VW Golf 2.0 automatico VW Golf 1.8 turbo auto

GMail users: please dogfood debug builds

For several months now, I've been seeing what appear to be DOM Window leaks when using GMail on trunk debug builds. The symptoms are that when the GMail tab has been open (and used) for a few days, memory leaks and the browser starts to feel sluggish. The sluggish feeling is caused by frequent cycle-collector runs that block the main thread for a short time every few seconds. I think it's related to GMail because the "--DOMWindow" output in the console shows that hundreds of DOMWindows (mostly with GMail URLs) are freed when I close the GMail tab. This is a very bad bug, but a

Useful read ?

http://www.tbray.org/ongoing/When/201x/2010/02/15/HTML5

pcnetsecurity@gmail.com =?UTF-8?B?QXNzaXN0w6puY2lhIFTDqWM=?= =?UTF-8?B?bmljYSAgbWFudXRlbsOnw6M=?= =?UTF-8?B?byBkZSBjb21wdXRhZG9y?= =?UTF-8?B?ZXMgaW5mb3JtYXRpY2Eg?= =?UTF-8?B?Vml0w7NyaWEtZXMgODI0NDU=?=

Contato: pcnetsecurity@gmail.com Contato: pcnetsecurity @ gmail.com Planos a partir de R$ 250,00 . Assist�ncia T�cnica Prestamos assist�ncia t�cnica nos computadores de sua empresa ou resid�ncia, e tamb�m possu�mos uma equipe qualificada para fazer a manuten��o no pr�prio local. - Contratos de Suporte e Manuten��o Reduza os custos de sua empresa com solicita��es de visitas t�cnicas para seus computadores, elaboramos um contrato de manuten��o integrado para sua empresa onde disponibilizamos: t�cnicos, equipamentos de suporte e substitui��o, e atendimento no hor�rio comercial ou

Scroll problems when using Mozilla Control

I'm using the Mozilla Control from XULRunner-1.9.2.en-US.win32 to embed a preview browser into my text editor application. I noticed with the 1.9.2 release that scrolling using the mouse wheel or using the keyboard doesn't work anymore. It used to work in v1.9.1. Can this be fixed?

Reverse thinking on extension architecture.

I have a suggestion to think about based on a js-engine discussion. How will extensions work if the ideas for partitioning the GC heap and multithreading being discussed on js-engine, http://groups.google.com/group/mozilla.dev.tech.js-engine/browse_thread/thread/7a173ebda62f32b1# move forward? As Shaver advocates, interconnecting threads via message passing (postMessage) has big advantages. Top of the list: multithreading acts like multiprocess acts like Internet multicomputing. Consequently we know how to deal with all the bad parts ;-). The critical ingredient in a success

Implementing the web timing spec

Hi everyone, there's a specification for exposing detailed timing information of web page elements via the DOM: http://dev.w3.org/2006/webapi/WebTiming/ Some people here would like to implement it in Firefox. The reason for exposing it via the DOM is to gather actual concrete end-user timings; other approaches don't allow that (also not all of this timing data is currently available to extensions). The idea is to change the spec a bit to limit the number of elements on which this data is exposed, probably to drop the Ticks interface, etc. Would people agree this is usefu

ASSERT: writer returned an error with non-zero writeCount

I have built FF 3.6 with DEBUG and I am trying to get Chromebug+Firebug to run on it without asserts. I assert here: --- nsresult rv = tee->mWriter(in, tee->mClosure, fromSegment, offset, count, writeCount); if (NS_FAILED(rv) || (*writeCount == 0)) { NS_ASSERTION((NS_FAILED(rv) ? (*writeCount == 0) : PR_TRUE), "writer returned an error with non-zero writeCount"); return rv; } --- in nsInputStreamTee.cpp. The reason is that mWriter is --- static NS_METHOD TestInputStream(nsIInputStream *inStr, void *closure,

Editable DOMs and HTML5 parser

I've noticed that test cases involving editability are failing with the HTML5 parser. Bugs on file: https://bugzilla.mozilla.org/show_bug.cgi?id=545405 https://bugzilla.mozilla.org/show_bug.cgi?id=540574 https://bugzilla.mozilla.org/show_bug.cgi?id=541078 * Editable stuff is editable. * Restoring editor state on navigation doesn't work reliably. * The editor doesn't inject its bogus <br> into the document. These look like the editor doesn't get attached to the document quite properly. Where should I look given that editing isn't completely broken with the HTML5 parser?

performance tests with extension-loaded profiles?

Hi, at some point I (or someone else) should work on https://bugzilla.mozilla.org/show_bug.cgi?id=525494, which would fix a few bugs in how the chrome registry finds the right locale codes. I suspect that that has TS impact, though probably only if there's a reasonable amount of extensions installed, the standalone builds with just one locale should hit a fastpath there. Anyone has experience in that? Axel

.def file

https://developer.mozilla.org/en/Compiling_The_npruntime_Sample_Plugin_in_Visual_Studio#Build In the step 9 and 10, what i do? I dont understand.

Mozilla 1.4 GTK2 - 1.7.x GTK2 and XULRunner 1.8.x - 1.9.0.x. ?

Where can one find Mozilla 1.4 GTK2 - 1.7.x GTK2 and XULRunner 1.8.x - 1.9.0.x. ? Am using GWT 1.5.3 and eclipse 3.4 on Ubuntu 9.0.4 ,however, I get the following error: ---------------------------- Unable to load required native library 'gwt-ll'. Detailed error: /home/smuwanga/software/applications/gwt-linux-1.5.3/libgwt-ll.so: / home/smuwanga/software/applications/gwt-linux-1.5.3/libgwt-ll.so: undefined symbol: JS_PropertyStub) ------------------------------ Someone ( http://www.eclipse.org/swt/faq.php#browserlinux )recommends installing Mozilla 1.4 GTK2 - 1.7.x GTK2 and XU

nsIPrincipal

The jetpack discussion inspired me to look back at evaluateInSandBox again, but so far my low opinion continues. In https://developer.mozilla.org/en/Components.utils.evalInSandbox is says: The sandbox must be initialized with either an origin URI or preferably a DOM window (nsIDOMWindow, like the window object in a web page). ...Using an nsIPrincipal is preferable to using an origin URI. I want a sandbox with system principal: how? I don't have a window (well hiddenWindow I guess). I don't have an origin. And I can't figure out how to create an nsIPrincipal. In

Changing URL-Input for Firefox-Browser

Hello, I am new in programming and I want you to ask how I can change in the source (where I can find in) the input (=URL-Bar) in the code. I want to change it in that way, that the browser receives the input from a input-field from a website (intranetsite). Can you please help me out? I hope I am in the right google-group :-) best regards

Web page components download order

I have created a following page. <html> <head> <title>Test page</title> </head> <body> &lt;script src="script1.js" type="text/javascript">&lt;/script> <img src="image1.png" /> &lt;script src="script2.js" type="text/javascript">&lt;/script> <img src="image2.png" /> &lt;script src="script3.js" type="text/javascript">&lt;/script> </body> </html> I would expect to see the download order of individual components as follows: script1.js image1.png script2.js image2.png script3.js But in fact it's (Firefox 3.6pre, order of HTTP-ON-MODIFY-REQUEST events): script1.js s

The technical basis for jetpack's restart-free feature

Mike Connor wrote: (on mozilla.dev.apps.firefox) ... > > For the investment in Jetpack to be worth it, for the users of our >product to gain the benefits of restart-free and sandboxed extensions, >we need to move as much existing development as possible to the new >model. I would like to understand the technical basis for the 'restart-free' feature. I attempted to copy this post to m.d.platform and set the followup there. Installing a new extension without restart seems pretty easy. What part of the current extension solution requires restart on first install? The

Certainty of https

From an extension I want to know that a page that claims to be eg "https://getfirebug.com" is certainly that page, not a phishing page. Is it sufficient to test window.location for host and protocol? jjb

The technical basis for faster updates with jetpack?

I've read about jetpack and looked at its code, but I still don't understand why some claim it will help Firefox cycle faster. The only thing I can figure is that API changes will be soaked up in jetpack, turning it in to a big converter box over time. Is there more to it than that? (Please I know jetpack has lots of things, I'm only asking about one. And yes I could ask jetpack newsgroup, I want to know what platform people think is the reason). Thanks, jjb

leaking memory when passing responses to Java-based observers - contractor sought

I'm looking to hire a contractor to help with some Mozilla work, specifically to get a Java application talking correctly to an embedded Firefox instance using javaxpcom/XPCOM/xulrunner. We are generating content in JavaScript, in a headless Firefox, and registering a java observer through javaxpcom. Somewhere along the way, the data being passed to the observer isn't being freed. I'm interested in contracting with someone to not only solve this specific problem, but also to provide some general knowledge to my team on embedding Firefox. Sorry if I've addressed the wrong list fo

Prefetch requests has no parent window.

A bug was reported in Firebug saying that the Net panel doesn't show prefetch requests. http://code.google.com/p/fbug/issues/detail?id=2459 The problem is that all the prefetched documents are requested without a relationship to the page that contained the ref="prefetch" (there is no parent window for the request). Since Firebug Net panel displays only requests that belong to the current page, these are skipped. Is this by design or I should file a bug about this? Honza

XPCOM Component Thread and JavaScript Callbacks

I am working on Firefox XPCOM Component (using C++) for Windows. My component has specific feature - it receives windows messages. I am using this approach: 1) Create a thread T in components constructor, 2) Create a window W in thread T (this window listens for messages), 3) Window W receives messages and sets some ("global") according received messages. So far so good. Now I want to call JavaScript callback from my component (as described here: https://developer.mozilla.org/en/Creating_JavaScript_callbacks_in_components). My callback is very simple function for testing reasons

Completion of background requests

When a XUL tree is used in a document, it causes "chrome://global/skin/tree/twisty-clsd.png" to be loaded. However, it seems to be loaded in the background, so nsIWebProgress::STATE_STOP events are fired for both document and window. What is the best way to observe background requests? -- Thanks in advance, Sergey Yanovich

nsIWebBrowserFind.findNext fails because there is no primary shell for the document

Hi, I am using nsIWebBrowserFind.findNext (from JS) and it always fails with NS_ERROR_INVALID_POINTER. Here is the code snippet that I am working with: // this.browser is the <browser> associated with a tab, got by gBrowser.getBrowserAtIndex(i) var browserFind = this.browser.docShell.getInterface (Ci.nsIWebBrowserFind); if(browserFind) { browserFind.entireWord = true; browserFind.matchCase = true; browserFind.searchString = this.pageRequest.search[i].content; contentMatchError = !browserFind.findNext(); } I tried debugging into the FF code and it turns out that this

DOMContentLoaded

The documention for DOMContentLoaded makes it sound quite insignificant: --- Fired on a Window object when a document's DOM content is finished loading, but unlike "load", does not wait until all images are loaded. --- Generally Firebug tries to get in to the page as early as possible. Based on the above, DOMContentLoaded does trigger some Firebug actions, but mostly minor ones. However I learned from Boris on Bug 536379 that events like DOMNodeInserted are not expect before DOMContentLoaded has fired. Are there other differences before that event? jjb

script blockers and similar techniques should be avoided

Script blockers (which were introduced in http://hg.mozilla.org/mozilla-central/rev/a05453237290 ) were introduced to prevent a specific type of serious bug that could be present in *uncommon* cases. However, the technique (which is also shared by the older IsSafeToFlush methods) has a serious risk of fixing bugs in one part of code only at the expense of causing them elsewhere. Because of this, I think we should avoid introducing new uses of this technique unless necessary (and definitely avoid introducing it to prevent problems in *common* cases), and we should try to reduce our e

<script> charset attribute ignored

An issue related to charset was reported in Firebug. If an html file is encoded in e.g. UTF-8 and included file encoded with different charest. The file is included as follows: &lt;script type="text/javascript" src="dirrerentlyEncodedScriptFile.js" charset="ISO-8859-1"/> The content of that JS file is not converted properly. Online test case: http://getfirebug.com/tests/issues/429/index.xhtml I checked the channel object (related to dirrerentlyEncodedScriptFile.js request) and it's contentCharset is not set. Accorging to the doc the contentCharset uses charset given in conte

suppressEvents accepts focus but not reload event

Recently we started calling suppressEvents when Firebug stops on a breakpoint. However this causes an anomaly: the DOMWindow accepts focus but not events. For example, control-R fails to reload but only if you happen to have focus on the now frozen window. Should this be a bug? jjb

Synchronous loading of about:blank

I'm looking at a bug where the problem seems to be that about:blank isn't being loaded synchronously into an iframe upon BindToTree but should be. So far I've found out that the start of the load is deferred until the end of the doc update (which still is "synchronous" as far as observability from &lt;script > goes), but after examining layers and layers of URI loading code, I haven't been able to locate what mechanism normally makes sure that about:blank loads synchronously. What's the normal mechanism for making about:blank loading synchronous and what might cause the mechani

About casting nsIDOM* interfaces in C++

Hello everybody! I'd like to ask a simple question about interfaces in XPCOM. I'm writing C++ XPCOM component for Firefox (linking against Gecko SDK 1.9.2 in MS Visual Studio). I am trying to cast from one interface to another one. Let's say from nsIDOMNode to nsIDOMHTMLElement. I retrieve initial nsIDOMNode from nsIDOMHTMLDocument via GetFirstChild (). Then I'd like to traverse DOM tree and cast each HTML element from nsIDOMNode to appropriate interface (nsIDOMHTMLElement, nsIDOMHTMLTableElement, ...). I check node name via GetNodeName() before calling QueryInterface(). Eve

DPI planning

Last week we had a meeting to talk about how to handle DPI and zooming in general. A consensus emerged that our old approach of automatically changing the number of device pixels per CSS pixel when the screen DPI is "too large" is not desirable. Instead, we should set that ratio based on system settings that correspond to user preferences, namely: -- On Mac, use the UI Resolution to set device-pixels-per-CSS-pixel (what Safari does) -- On Windows, use LOGPIXELSY (what IE8 does) -- On GTK/X, use Xftdpi We should step up our support for physical units in CSS by correctly mea

Adding test-only APIs

We want to add tests for cross-platform IME infrastructure. These tests depend C++ APIs that aren't currently scriptable, but we'd like to write tests using script because C++ is a bit cumbersome, especially for working with complex documents. In other situations like this we've exposed API through nsIDOMWindowUtils, but that's suboptimal because we ship it, and I don't really want to ship these test-only APIs. Any suggestions? Rob

Proposal for a different terminology for windows.

One of the biggest bugs in the mozilla platform API is the overloading of 'window'. Enumerable conversations stumble around as the participants try to agree on what each person means by 'window'. Lots of bugs, some very subtle and long standing, are caused by misunderstandings about the meaning of 'window'. Some specific examples: https://developer.mozilla.org/en/SpiderMonkey/Split_object https://bugzilla.mozilla.org/show_bug.cgi?id=534149 https://developer.mozilla.org/En/Working_with_windows_in_chrome_code The problem here is that 'window' is not an abstraction in most of the

mozilla-central is open again

Hi everyone, We're down to the last few bugs for mozilla1.9.2/Firefox 3.6, and = they're localized enough that we've decided to re-open the = mozilla-central tree for checkins. Since we expect that there might be a = backlog, Ted's (been) volunteered to help co-ordinate landings today. = He's in #developers. cheers, mike=

FF 3.6 vs FF 3.5 and use of nsIDOMWindows for error pages.

I hit a weird problem in Firebug and I've only come up with one equally weird possible reason for it. If I run Firebug 1.5b6 on Firefox 3.6 with the home page file:///C:/DOCUME~1/JOHNJ~1.BAR/LOCALS~1/Temp/issue-420.html which does not exist, I get an error page. To see the issue easily, now select the Script panel. Next load the url (actually I think any valid url will work) http://getfirebug.com/tests/content/script/singleStepping/index.html The page loads, but Firebug's script panel still shows the previous url. There are a lot of ways Firebug could have this bug. But even

100% browser CPU.

The page https://developer.mozilla.org/en/dom/document.getelementsbyclassname does not document the method call return types, so I thought I would just check the link at the bottom of that page: http://whatwg.org/specs/web-apps/current-work/#getelementsbyclassname Unfortunately my CPU pegged 100%. Naturally I thought Firebug has jammed things up, but I saw no trace outputs. In fact, I could get almost nothing out of Firebug/Chromebug because the CPU was pegged. Eventually I tried in a clean profile and the CPU is still pegged. While looking through bugzilla on another firefox i

Get an elements specified style

In the following document, I have an absolutely positioned element with id of "b" that was positioned from the right. How can I tell how it was positioned? If I get the computedStyle I get both left and right with no knowledge of which one is fixed (that is, if I resize the window, the value of right will stay the same, the value of left will change). The only way I can think of is to iterate through all the styles that apply to this object from getCSSStyleRules() to find what the collapsed specified style is. Is this correct? Is there a shorter/faster way to find the specified style?

Time to require SSE2

I think this is a reasonable requirement for releases after Firefox 3.6. Does anyone disagree?

Fwd: how browsers transform URLs

Forwarding to .dom (because there are recommendations for script interfaces, which Erik claims are particularly inconsistent) and to ..platform. Apologies if either or both is the wrong place. Gerv -------- Original Message -------- Subject: how browsers transform URLs Date: Wed, 25 Nov 2009 11:49:36 -0800 From: Erik van der Poel <erikv@google.com> We are happy to announce the open source release of Client URL Internet Emission Sniffer (CURLIES). The purpose of this project is to see how browsers and other Web clients transform URLs as they access them. This is done by g

Application update using SSL and own certificate

Hi, I have another problem with the application update. Using an HTTP app.update.url works without problems. The problem comes with HTTPS URLs for app.update.url. I add our in-house certificate using JavaScript with the method described here: https://developer.mozilla.org/en/Code_snippets/Miscellaneous#Adding_custom_certificates_to_a_XULRunner_application This works great for normal XMLHttpRequests, I do them all the time. But doing an "check for updates" gives me the catch-all error "xml malformed (200)". A debug build and the highest possible NSPR logging gives me a huge l

Checking if an nsIDocument is being examined by the user

nsContentSink alters its flushing depending on queued user events on the event loop. I was thinking of implementing a similar idea for the HTML5 parser. How can I tell if an nsIDocument is being displayed in the frontmost tab (perhaps in a frame inside the frontmost tab)? Is there a way (nicer than what nsContentSink does) to detect if the user is actively interacting with the frontmost tab? -- Henri Sivonen hsivonen@iki.fi http://hsivonen.iki.fi/

how to get coordinates of particular word or letter in a text node

hi, I'm using nsIDOMNSElement.getBoundingClientRect() to get coordinates of a DOM node. But I'd like to get coordinates of particular word or letter in a text node. I guess I could calculate this by myself if I only knew the node's text line height, letters widths and line wrapping. Is there any API for that? thanks Jacek

XULRunner NSView

The wiki page says the following : NSView-based-widget (Mac OS X only) (not yet complete) I'm wondering what progress has been mad on this since the wiki page was updated two years ago... Is it yet possible to embed Gecko or XULRunner in a Cocoa app in the form of an NSView (or any other way, for that matter)? Thanks.