Talk:Acid3/Archive 2

This is an archive page for Talk:Acid3

What is being tested?
I have changed the article to reflect this sentiment: "Acid3 is primarily concerned with ECMAScript and the DOM though Selectors Level 3, Media Queries, and data: URIs are also tested." There is very little CSS and SVG in this test.

From Ian's comments: "there are 6 buckets with 16 tests each, plus four special tests (0, 97, 98, and 99)"


 * Update: This quote is now gone from the source code. It's still true, though.--itpastorn (talk) 13:30, 16 February 2008 (UTC)


 * Bucket 1: DOM Traversal, DOM Range, HTTP
 * Bucket 2: DOM2 Core and DOM2 Events
 * Bucket 3: DOM2 Views, DOM2 Style, and Selectors (CSS 3 selectors)
 * Bucket 4: HTML and the DOM
 * Bucket 5: Tests from the Acid3 Competition (contribued by the "public"; SVG, HTML, Unicode...)
 * Bucket 6: ECMAScript

Test 99 "The weirdest bug ever"

--itpastorn (talk) 11:35, 6 February 2008 (UTC) Modified: --itpastorn (talk) 13:30, 9 February 2008 (UTC)
 * Firefox tracking bug
 * Firefox tracking table
 * Webkit tracking bug
 * Web Standards Project Announcement

According to the announcement from The Web Standards Project, the following is being tested: Tests added by the open competition, including some on SVG, is not listed by WaSP. --itpastorn (talk) 13:36, 9 February 2008 (UTC)
 * DOM2 Core
 * DOM2 Events
 * DOM2 HTML
 * DOM2 Range
 * DOM2 Style (getComputedStyle, …)
 * DOM2 Traversal (NodeIterator, TreeWalker)
 * DOM2 Views (defaultView)
 * ECMAScript
 * HTML4 (, , …)
 * HTTP (Content-Type, 404, …)
 * Media Queries
 * Selectors (:lang, :nth-child, combinators, dynamic changes, …)
 * XHTML 1.0
 * CSS2 (@font-face)
 * CSS2.1 (’inline-block’, ‘pre-wrap’, parsing…)
 * CSS3 Color (rgba, hsla, …)
 * CSS3 UI (’cursor’)
 * data: URIs

Even more links about what is being tested: --itpastorn (talk) 13:54, 9 February 2008 (UTC)
 * John Resig - Acid 3 Tackles ECMAScript
 * SVG added through the open competition (From the guy who contributed most of them...) See also this message on the SVG mailing list and this one.

I am adding this link about the Ahem font for future reference as well. It also features prominently in MS CSS 2.1 tests. And we really should have references to the formal test suites as well.--itpastorn (talk) 18:46, 8 March 2008 (UTC)

The test for the downloaded font is the white 'X' with a fuchsia background. If the font is downloaded the it will not be visible as the font being used is the Ahem test font -- the glyph for the X character is a rectangle with no padding, so correctly rendered it will be a white rectangle in the upper right corner of the test, on a white background so is invisible when rendered correctly. I have tested the webkit nightlies, but am unable to test IE. -- 24.130.131.58 (talk) 06:15, 9 March 2008 (UTC) (Moved to this section.)--itpastorn (talk) 12:28, 9 March 2008 (UTC)

Linkdump: CSS selectors tested --itpastorn (talk) 12:24, 29 March 2008 (UTC)

Linkdump again: Ian explains the "smoothness" criteria in more detail

Not a test suite
This is from a private mail to me from Ian H: "Acid tests aren't test suites, they're demo pages that attempt to exercise many aspects of standards in order to encourage people to pressure browser vendors to consider standards-compliance a priority. If browser vendors made test suites and fixed the bugs that the test suites found, then they  wouldn't have any bugs left for Acid tests to pick on." This means that both this article and Acid2 really should be re-worded.--itpastorn (talk) 10:38, 3 March 2008 (UTC)
 * This is exactly what a test suite is. It just doesn't test all features of the specs tested. The nuance about the purpose and the targeted public doesn't change the nature of a test. --Fenring (talk) 11:56, 3 March 2008 (UTC)
 * A test suite is a collection of individual standalone tests. Acid tests are not. Annevk (talk) 16:38, 5 March 2008 (UTC)
 * On the sites http://www.webstandards.org/ and http://www.acidtests.org/ Acid2 or Acid3 are consistently called "test page(s)" or simply "test(s)", never "suite(s)". So, does anyone have a more authoritative source than that?--itpastorn (talk) 09:17, 13 March 2008 (UTC)
 * But they meet the definition of Test Suites, don't they? ff m  19:02, 13 March 2008 (UTC)
 * Nope! One might tentatively argue that Acid3 is a test harness, but it is not a suite. BTW, considering the discussion on the Acid2 page about my perceived OR when I deduced through simple reasoning (and provided resources) that IE 8 beta 1 fails. I have sourced the use of the word page. Who can source the use of the word suite?--itpastorn (talk) 21:26, 13 March 2008 (UTC)
 * Synthesizing material to reach a conclusion certainly is original research. On the other hand, using simple terms in a way that everyone agrees on is not. What do you mean by "source" the use of words? Just use words. -- Schapel (talk) 21:38, 13 March 2008 (UTC)

Shift+click A
Should this information be in the article? It has now been removed twice.

Personally I think it makes sense, as opening the results in a new tab to me is preferable to opening them in a JavaScrip alert box. The fact that it is possible to see the results by clicking the A is not obvious in any way, unless you accidentally hover over the letter. That you get a different result from clicking while holding down shift is even harder to know.

Maybe one could rephrase the instruction, but I do think it is of value to the reader.--itpastorn (talk) 08:21, 5 March 2008 (UTC)

Betas
Not including information on beta releases is a great way to artifically reduce the quality and usefulness of this article. This is not a good goal -- think about what our readers would hope to find when visiting Wikipedia; they want to know the latest information that we can get from reliable sources. This includes the current development state of major browsers, especially when it's something like Internet Explorer 8. -/- Warren 00:23, 6 March 2008 (UTC)
 * First of all it is a matter of WP policy. In what way is it not original research? I have proposed a solution on this very talk page, where we could have the information about "compliance efforts" (good catch renaming the section, BTW), written in such a way that it is of encyclopedic value, and not recentism.--itpastorn (talk) 10:47, 6 March 2008 (UTC)


 * You probably shouldn't be trying to use the "recentism" argument on an article that describes something that was officially announced just three days ago. Also, you are misunderstanding the thrust of the recentism essay.  The goal is to dissuade people from adding information that isn't important or doesn't have long-term value.  In the context of Acid3, the state of browser support is absolutely vital to the topic, because it is very much a current issue, one that our readers will be interested in tracking.  Also, original research is effectively countered by reliable sources, and in the case of software, OR arguments are weakened by the use of screenshots.  -/- Warren 13:02, 6 March 2008 (UTC)


 * I believe that including the latest information with regards to preview releases and betas is very important to the article. Showing the differences between version X and version Y of the same browser gives the reader a perspective on the difficulty and progress of implementing standards compliance in browsers. Because browser standards implementation develops rapidly compared to other types of software (e.g. Word Processing), it is especially important to have the most up-to-date information. Secondly, people who come to this web page will be looking at which browser performs best, perhaps to make an informed decision about which browser to use, or which browser to test their web content in. Due to the nature of the Acid3 test (it was explicitly designed so that none of the current browsers pass the test) I believe it is very important to show the renderings of upcoming browsers. - ARC Gritt TALK 13:20, 6 March 2008 (UTC)


 * Nightlies are unstable, unfinished softwares with experimental features and regressions. These are definitely not long-term informations. If you consider that the percentages of passed test is an important information, why don't you provide more details about e.g which subtests are passed and which are not ? This is crazy. Who needs that ? It's not encyclopedic. Though, a notable information that will be in the article forever is the first engine (beta or not) to pass the test (it means fully). Look at the Acid2 article. It's much more informative than these everchanging values. And Warren, the inclusion of this article is not recentism. It is there to stay. It is very persistent information in web history. --Fenring (talk) 15:53, 6 March 2008 (UTC)
 * Look at how ACID2 does it, and perhaps copy from there? I do think noting the status of the various engines at the release of the test could be interesting and useful. IE7 had 17 points as of the release, safari has 33 points, etc. That gives readers an idea of how things started out with the official test. I will say regardless, this test is quite important in the world of web programming and browser development. All major browsers still don't pass ACID2 in their release versions, still waiting on IE8 here. :) I'm just commenting here because I read the article. LinEagle (talk) 23:13, 8 March 2008 (UTC)
 * The Acid2 article shows when each browser passed for "compliant" applications and screenshots for "non-compliant" applications. Because no browser passes Acid3, that means we're left with screenshots for each browser until some browsers pass. The score at time of release has no purpose, because many browsers (especially Safari and also Firefox) have been specifically fixing bugs that lower their score for months, so they do not how how each browser "started out". -- Schapel (talk) 00:24, 9 March 2008 (UTC)
 * Well, of encyclopedic value is, IMO, not so much the scores per se as how this test has caused vendors to improve their products. I can do without the scores at the time of the release per se. But what happened from the date Anne V K first told the world about the test until it was released? What has happened since? Since this article is about the test, its details and purpose, and its impact - not about browsers! - this should be the focus of the article. And I note that apart from small grammar fixes so far I am the only one who actually is contributing substance to the article in this regard. That's the reason I removed the screenshot with Webkit scoring 90 and reinstated the screenshot where webkit scored 88. It illustrates the text better.--itpastorn (talk) 15:34, 11 March 2008 (UTC)

The new table
Kudos to Schapel for his work in getting the table together. It looks good. I have one comment, though... there's no information as to which release is referenced by the screenshots. Could this be incorporated somehow? -/- Warren 19:47, 13 March 2008 (UTC)

CSS is a technology, described in a spec that may be at a draft stage. Check the W3C standardization process.
It is factually wrong to call downloadable fonts and text-shadow drafts. They are CSS technologies described in drafts.

It is factually wrong to say that Acid3 requires support for the drafts in which these technologies are mentioned. The specs contain many things not in the test.

CSS is a technology, a declarative markup language to be more exact, described in specs that may be at a draft stage. Check the W3C standardization process.

Also, if you look at the specs, almost every CSS spec is at the draft stage. However, individual items from those specs are at a very stable state. Therefore it is actually somewhat misleading to call downloadable fonts and text-shadow drafts (they are not mentioned in the Beijing document, but have been on the CSS mailing list).

You may also see that some specs are about to be demoted from the CR to the WD stage. And the only recommendations that are finished (1.0 and 2.0) have been declared obsolete.

And yes, the W3C also call their languages technologies: W3C technology stack --itpastorn (talk) 23:25, 13 March 2008 (UTC)


 * Wow, that's a big pile of steaming........ stretch.


 * I have a novel idea for you: RESEARCH YOUR CLAIMS.  Do it.  Don't fucking argue.  Don't fucking edit war.  Don't fucking engage in original research.  None of those things are welcome on Wikipedia.  An  utterly simple Google search on the term "CSS Technology" doesn't return anything interesting about Cascading Syle Sheets, and only about 30,000 results total. ("css technologies" returns less than 3,000 hits).  The W3C does not describe CSS as a "technology".  I will repeat that in bold letters: The W3C does not describe CSS as a "technology".  The word does not appear in the specification!  This means that we do not describe it as a technology, either.  This is not up for debate or discussion.  As a Wikipedian who is bound by WP:V, WP:OR and WP:RS, you absolutely do not reserve the right to invent terminology, or to promote rarely-used terms.  CSS is a specification, not a technology.  It says so right in the title.  CSS is a standard, not a technology.  CSS is a language, not a technology.


 * Now before your male ego gets all fired up for an argument because I've just told you that you're wrong... STOP. THINK.  RESEARCH.  Do you want to do this using the valid, authoritative information we have available to us, or not?  CSS3-text clearly identifies itself as a working draft.  CSS3-webfonts clearly identifies itself as a working draft.  The beijing document is completely irrelevant because it doesn't mention CSS3-text and CSS3-webfonts.  Acid3 tests things that are in the CSS3-text and CSS3-webfonts documents, and doesn't mention the beijing document at all.  Acid3 uses things that are declared in draft standards.  That's the bottom line.  You're wrong -- don't fight this, you won't win.  -/- Warren 23:50, 13 March 2008 (UTC)

Third Opinion:
 * Guys, calm down.
 * Warren, Please be WP:CIVIL. User:Itpastorn hasn't swore, yelled (IN CAPS), etc. He just reverted you, once.
 * Itpastorn: they are actualy "W3C Working Drafts". However, they are drafts of technologies. You can call them both, or just go with W3C Working Drafts of the CSS standard. ff m  00:14, 14 March 2008 (UTC)


 * Two questions. Yes or no, A or B. (1) Is CSS part of what W3C calls the "technology stack"? (2) Is the word "draft" used by the W3C to denote a document at a certain stage (A) or something described within such a document (B) according to the described standardization process?


 * BTW, the Beijing document was not introduced to support the claim that downloadable fonts and text-shadow per se have passed the draft stage. It was an example about how the W3C CSS Working group are trying to illustrate the principle that some parts of CSS 3 are implementable even if they are described in specs that as whole still are at the draft stage. I did not propose we put any info about the current readiness of a certain CSS technology in the article. I have no inclination to wade through the mailing list archives and other discussions from the W3C Working Group. I proposed we'd remove it, as it is quite irrelevant to the article - and when badly worded also misleading.


 * @Firefoxman. Thanks for your admonition. I am actually very calm as I have researched this issue by reading quite a lot of W3C documents in my day, as well as having participated on the CSS WG public mailing list and the WHATWG mailing list for quite some time now, and I am not shaken at all by Warren's screaming and cursing. I know my W3C terminology.

--itpastorn (talk) 00:48, 14 March 2008 (UTC)


 * I don't think what W3C does with CSS can be called "technology". It just says how the markup is to be interpreted. The "technology" part of realizing that is left to the browser makers. I think a better working would be "...downloadable fonts and text shadow specifications, which are under consideration by W3C for standardization". And those who are not comfortable with the word "standardization" (W3C isn't a standards body, it cannot propose standards; it can only do recommendations), replace it with "recommendation"-derived word of your choice. --soum talk 05:29, 14 March 2008 (UTC)
 * The wording is fine by me. Go ahead, make the edit.--itpastorn (talk) 08:52, 14 March 2008 (UTC)


 * I made the edits along with a couple of other changes. Feel free to modify it as necessary. One thing I would like to point out: "In order to render the page with the required pixel perfection..." - the rendering cannot be pixel perfect with the reference rendering, nor will it be identical across browsers as the browsers handle text rendering differently, especially subpixel rendering. So, its better not to mention being "pixel-perfect". --soum talk 10:51, 14 March 2008 (UTC)
 * Soum, I think on the same browser/platform, the test and the reference rendering must be pixel-identical. Even if the reference rendering is different across different browsers. --Fenring (talk) 14:59, 14 March 2008 (UTC)


 * I am sorry I am not getting you. Are you implying there are multiple reference renderings, one for each OS/browser combo? --soum talk 15:08, 14 March 2008 (UTC)
 * I am. If you look at the source code of the reference rendering page, you'll notice it's not a raster image, but html/css instead (that uses simple features implemented by every current UA). The rendering must be the same with this simple way and the complex-test way, for one user-agent. But it not required to render the same across browsers. --Fenring (talk) 15:24, 14 March 2008 (UTC)


 * I stand corrected. Thanks. --soum talk 15:31, 14 March 2008 (UTC)

A few more notes about sources
Since I believe myself to be the one who has contributed most of the factual content about what is being tested, how the "buckets" work and get their color, etc, I think I should explain my sources a bit more clearly. Not every claim is validated with a footnote right now. My main source is the source code of the test, which means that it is verifiable, but not in a simple fashion - even though Ian has put many helpful comments in the source (I've studied it thoroughly). When WaSP releases its companion guide we will get a better resource. Until that time comes did I however believe that people might want to know more about the test than what the current score is for browser X. Footnotes will be provided in due time!

The test does not test compliance with the Web Fonts draft, as it is not a full test suite. My reference was not the full draft but one item in it. For example, Acid3 does not test "font-stretch" or "unicode-range". Downloadable fonts is however a part of this spec that is in very high demand and therefore is unofficially declared implementable. I would not dream of saying that in the article, as it would take me a long time to dig through the mailing list and bug databases in order to source it. But this is a talk page and I want to explain my thinking.

Similarly, the test does not test for compliance with the CSS Text Level 3 draft. Hyphenation is one example of what is not being tested. "Punctuation-trim" is another. Whitespace, word-spacing and letter-spacing, OTOH, are in. And the much longed for text-shadow. Once again my reference was to an individual item in the draft.

I happen to know that downloadable fonts and text-shadow is in high demand by designers, and it is factually correct that Acid3 checks for it. I could have mentioned letter-spacing, but that topic is not so hot. Therefore I did not prioritize it. BTW, want to know why FFox 2 does not render the buckets as squares, but as lines? It does not support "inline-block"! (Line 36, column 33 in the source code.)

If you think that all of this is OR, by all means feel free to remove the entire section from the article.--itpastorn (talk) 01:38, 14 March 2008 (UTC)


 * In the absence of any reliable source that contradicts what's in the article, just put everywhere a citation should be. Someone finds a reliable source that says something different, they can remove your OR, and if someone challenges your OR after it's been marked as needing a citation for a long time, they can remove it. But we shouldn't remove it immediately just because it's OR. -- Schapel (talk) 11:30, 14 March 2008 (UTC)

Current events
Perhaps we should stick this template into the article above the table with the scores:

BTW, Webkit is at 91, but since they have no working code at all for SMIL that i know of, I suspect it's going to be a while for them to reach 100.--itpastorn (talk) 21:53, 15 March 2008 (UTC)


 * The test has been changed slightly. It appears that most browsers will get an automate 1 point bump in their scores. I don't know how to best add this news to the article so I'm posting it here. http://webkit.org/blog/161/webkit-hits-93100-in-acid3/ —Preceding unsigned comment added by Pattersonc (talk • contribs) 01:24, 18 March 2008 (UTC)

Yeah, I get 53/100 with FF2 now. —Preceding unsigned comment added by 81.182.143.92 (talk) 09:36, 19 March 2008 (UTC)
 * When are we going to replace the screenshots of the test rendering to accoint for the 1 pt. bump? --soum talk 17:06, 23 March 2008 (UTC)


 * No one objected. So I updated the IE8 image. --soum talk 15:26, 25 March 2008 (UTC)

AcidAhem downloadable font
I first put the info in, then I thought I was wrong and now Anne VK has changed it back to my initial understanding. These are the two lines of code (I split the 2nd one up here to avoid scrolling):

@font-face { font-family: "AcidAhemTest"; src: url(font.ttf); } map::after { position: absolute; top: 17px; left: 639px; content: "X"; background: fuchsia; color: white; font: 20px/1 AcidAhemTest; }

It's line 25 and 26 in the source code of the test.

17 pixels down and 639 pixels to the right. That can't be the purple X (seen above the instructions on the FFox screenshot - right?) But it's still "background: fuchsia". That can't be the red square in the upper right corner - right?

So looking at the screenshot in Opera 9.26 it is in a more reasonable position, almost at the top.

I conclude that Gecko puts this at the wrong place, but it is indeed the white X on fuchsia background that is not supposed to be seen. Buggy browser however might not put the glyph in the correct position. Now I need a source besides the source code of the page - and a good way to phrase the article text...--itpastorn (talk) 16:41, 18 March 2008 (UTC)

Webkit cheating?
Got myself linked to this trac, and it looks like webkit is doing a special exception incase it is the acid 3 font in order to pass the test. Source: http://trac.webkit.org/projects/webkit/changeset/31322 I refrained from editing before looking into the matter more carefully. Lyml (talk) 14:47, 27 March 2008 (UTC)


 * Mozilla developer Mike Shaver says Safari is treating the Ahem font as a special case: http://shaver.off.net/diary/2008/03/27/the-missed-opportunity-of-acid-3/ This, of course, defeats the purpose of the Acid3 test. -- Schapel (talk) 16:02, 27 March 2008 (UTC)


 * Yes, that change is quite specific to that font, which of course is generally frowned upon (hardcoding something just to beat an internet standards test doesn't improve anything). However, you should also look at the related bug for this to see both sides: . nneonneo (talk) 19:31, 27 March 2008 (UTC)


 * At least this act could be cited as proof to support the notion that Apple/Webkit really wanted the PR of winning this race! Personally I also believe people should make a clear distinction between commitment to standards and commitment to winning the Acid3 race per se. In my comment on Mike Shavers blog I have provided a strong arguments (IMHO) that Mozilla might actually display a higher commitment to standards - by postponing some Acid3 fixes! see also this comment on Mozilla Bugzilla


 * What is of encyclopedic value, though, is what kind of techniques that developers can use and users enjoy, once the technologies in Acid3 have widespread support from stable browsers. (And if a dev is using progressive enhancement, widespread does not mean wait for IE...) Putting such stuff into the article is way more productive than scores or discussing if a browser cheats or not.--itpastorn (talk) 10:43, 28 March 2008 (UTC)


 * However the techniques the developers can use are embedded fonts with the name 'Ahem' if you want them to act correctly. This is just silly and should atleast get a mention in the 'pixel perfect' remarks. Which wouldn't be pixel perfect if the test was exactly the same except the font name was changed to something else than 'Ahem'. It's not really fair to call webkit xxx the first browser to render acid3 test pixelperfectly if all they did was make a special hardcoded exception for the test (even though it was just for font rendering in this case, and not for the entire test suite). Well I'll still refrain from editing (as this will surely cause an edit war where people are arguing over silly things like who came first). I don't beleive it is relevant, however should it be noted, it should atleast be the thruth. Lyml (talk) 23:26, 28 March 2008 (UTC)


 * Hixie is now changing the test and the WebKit team will back-out the hardcoded behavior . So, the question isn't relevant anymore from an encyclopedic point of view.


 * From an encyclopedic point of view, that means WebKit only has a public release that truly passes the "pixel-for-pixel" part of the test in 29 March, one day later than Opera's public release of their WinGogi build.Ufopedia (talk) 04:57, 4 April 2008 (UTC)


 * The change means that WebKit r31356 no longer is pixel-perfect. Tim Altman (talk) 20:58, 31 March 2008 (UTC)


 * Do you have a source for these claims? It doesn't make sense to me, as moving the graphics the way they did should mean WebKit would render the test correctly whether the Ahem font was antialiased or not. Besides, it was an issue only on Macs, and WebKit also runs on Windows (Opera tellingly did not release a Mac version of WinGogi that renders the test correctly). From an encyclopedic point of view, all that matters is that by the end of March, there were early development versions of Presto and WebKit that score 100/100 and render the test correctly. -- Schapel (talk) 13:34, 4 April 2008 (UTC)

Timeline of compliant layout engines
I think this section should remain commented out until at least one browser actually passes the Acid3 test. We already have better information about browsers that come close in the previous section. -- Schapel (talk) 18:46, 27 March 2008 (UTC)


 * Agreed. There's no point in calling them "partially compliant" -- they either are, or aren't (and all browsers currently released fall into the latter category, especially because of test 26, the performance test). (by the way, I hit enter accidentally while entering the edit summary for ; if anyone was curious, I meant to say "Restore comment (no point in saying they are partially compliant)") nneonneo (talk) 19:34, 27 March 2008 (UTC)

Actually, having a timeline at all emphasizes the "race" aspect that has nothing to do with what the Acid3 test is about. This race is just leading to anonymous editors coming in and adding inaccurate information to this article. It doesn't really matter when the layout engines that pass were released. I propose simply listing the layout engines that pass after a stable release that passes is released. -- Schapel (talk) 14:14, 29 March 2008 (UTC)


 * I see no way of stopping that. Sigh! We just have to revert it. As for Webkit, until the tracking bug is closed or we get an official announcement from dave Hyatt we just have to keep saying any talk of Webkit passing is just a rumor. Right now they are opening more bugs about the glitches related to caching and JavaScript speed. I am quite sure Opera will issue official statements as well. And then we have the issue of Ian saying that he is still not sure exactly what speed measurement that should be used to monitor the third criteria. I.e. the test itself is still not completely finalized!--itpastorn (talk) 19:12, 29 March 2008 (UTC)


 * I also think we should eliminate the "browser competition" section. The Acid3 is not a competition. It has nothing to do with which browser does best or which one passes first. I think we should remove all dates from events that don't have to do with the test itself. Stating the current status of the latest stable and development builds is the most important information. -- Schapel (talk) 04:42, 30 March 2008 (UTC)


 * Yep, we should. It was never developed as a competition, and getting to compliance first doesn't achieve anything until all major browsers pass. Plus how can it be a competition when both MS and FF are out of the race? I am removing it. --soum talk 05:24, 30 March 2008 (UTC)


 * What it was meant to be, and what it actually did are not the same thing. Sure it was not meant to cause a race. But it did, and I see no reason to hide that. Ariel. (talk) 14:51, 2 April 2008 (UTC)


 * Agreed. That there was a race between Opera and Safari developers is notable. That there were public releases of WekKit and Presto that scored 100/100 and rendered the test correctly by the end of March 2008 is notable. A timeline of events or who did what first, however, is not notable. -- Schapel (talk) 15:14, 2 April 2008 (UTC)

Instead of a timeline (which makes it seem like Acid3 is a race), I propose we list the first stable versions of browsers and other programs that pass Acid3. This would not only clarify which versions of which programs pass Acid3, but would also emphasize that having a layout engine that passes Acid3 doesn't do much good until programs that use it are released for general use. Perhaps it could look like this: The name of the section could be Compliant layout engines and programs. -- Schapel (talk) 13:16, 12 April 2008 (UTC)

Updates needed
The current nightly build of WebKit passes the Acid3 test 100%. I think we should update the article to reflect this. --156.111.38.218 (talk) 23:16, 30 March 2008 (UTC)

Sorry, this has already been done. Thanks. --156.111.38.218 (talk) 23:21, 30 March 2008 (UTC)

WebKit doesn't pass. They still have performance issues. See http://bugs.webkit.org/show_bug.cgi?id=17510. In other news, the WebKit section says, "WebKit build r31356 scores 100/100 and produces a correct rendering", but that's not true. On March 29th, Ian Hickson updated the test again (http://ln.hixie.ch/?start=1206756775&count=1) to change anti-aliasing. This caused older WebKit builds to have rendering problems. Tim Altman (talk) 00:10, 31 March 2008 (UTC)

Opera scored 106 with developing build —Preceding unsigned comment added by MartinZM (talk • contribs) 04:07, 3 April 2008 (UTC)

Wii Internet Channel
It may be interesting to note that the Acid3 test not only fails on the Wii's Internet Channel, but it completely freezes the browser...as well as the entire system. Its effect is so profound as to force the user to turn off the Wii using the Power button on the console itself, or unplug it.

Just throwing that out there... Kewlio (talk) 07:03, 1 April 2008 (UTC)

David Baron
http://dbaron.org/log/20080406-acid3

Do you think this Mozilla Dev's op-ed about his plans to implement to the test are relevant? It might be problematic in that it would be more difficult to get responses from devs of some of the other layout engines. gren グレン 10:18, 9 April 2008 (UTC)


 * After reading the piece, it looked to me that the most useful piece of information is debunking the idea that the browsers that score 100/100 do so because they support more standards. I think the article needs more information on what the purpose of the test is (to me, it is about browser developers cooperating, not competing) and how to interpret the score of the test (the score does not reflect general standards support, but merely how well browsers do on what the Acid3 test tests). -- Schapel (talk) 12:16, 9 April 2008 (UTC)


 * If one follows the movement on Bugzilla for Gecko on bugs related to Acid3, there is movement on three issues: Text shadow - probably not enough stability to make int FFox 3, CSS 3 selectors - might make it into FFox 3 and media queries - which again probably won't. I think it is reasonably clear that these issues have received a slightly higher priority thanks to Acid3. However, SMIL is still not on the radar, and for sure will not be implemented just to pass the test. (Webkit is continuing to improve its SMIL support, even though they have "enough" to pass the test. I am following their bugzilla too.) What David baron has done is to take open issues about CSS selectors to the CSS WG and discuss them, while implementing, delaying Acid3 performance in favor of a better Standard! I think this is one prime example of how commitment to standards meanS (A) not (B) yet passing Acid3. (This can be followed on the CSS WG issues list, Bugzilla and the CSS WG mailing list.)


 * But today I googled a bit on Acid3 to check for new sources. The following stood out: (1) Almost everyone writes about it as if it's a race. (2) Almost everyone who writes about, except for official sources, have really no clue about what is being tested. The Guardian even claimed it was a test of the "latest" HTML standards" from W3C! (3) The smoothness criteria is really hard to understand. Even Roger Johansson and Robert Nyman got it wrong and claimed that Webkit and Opera passed the test. I actually wrote to Ian and suggested that he add a visual clue when the "smoothenss" passes or fails. His blog post about what it really means is basically his reply to me. Then came Kimberly Blessing - co-lead of WaSP and said the same thing (sic!), only to be debunked by people from opera and Webkit (in the comments below her post)!


 * I have planned for a long time to actually write more about the test as such, and what it means to developers. But as of now any such writing will be original research.--itpastorn (talk) 15:45, 10 April 2008 (UTC)


 * Perhaps we should start with the Acid2 test and try to explain that its purpose was actually a challenge to Microsoft to support standards in IE. Of course, it was a challenge to other browsers as well, but the point was not a race or competition, but an effort to get developers of all browsers (especially IE) to cooperate and be interoperable by supporting standards. Acid3 is exactly the same, although we will need to cite a source to specifically state that this same thinking applies to the Acid3 test also. We should try to make it clear that how soon non-IE browsers pass the test is unimportant. How fast Microsoft fixes IE is what's important, as it seems to always be fixed last, it has more users than other browsers, and users of old versions of IE seem to upgrade to the latest versions more slowly than users of other browsers. -- Schapel (talk) 15:24, 11 April 2008 (UTC)

April Acid3 changes
I don't think the latest updates on Acid3 changes are correct. Safari 3.1 still scores 75/100, both when I try it, and also on Browsershots. Opera 9.5 beta 2 scores 78/100 on Browsershots and also according to a forum post and a foreign article. From Hixie's description, likely any browser that scored 100/100 before now scores 99/100, because one test was changed. Let's just mention that Hixie changed the test, and this means that the 100/100 scores reported previously may not be valid, as there was an error in the test. Both pieces of information are correct and verifiable. -- Schapel (talk) 12:19, 25 April 2008 (UTC)
 * Firefox 3.1.preAlpha 81/100--144.122.250.92 (talk) 11:08, 15 July 2008 (UTC)

WebKit r30885 at 88/100 picture?
What's the point of this picture now? Like what the text says, it's already way outdated, and I don't think it serves any encyclopedic purpose now. 61.51.141.65 (talk) 14:46, 3 June 2008 (UTC)
 * It serves as an example screenshot to demonstrate the meaning of the various shades of grey and colors. I agree it should be updated. A screenshot of Acid3 in Opera 9.5 seems like it would be the most appropriate, because it's the most used browser that exhibits both shades of grey and a color. Why not grab a screenshot from browsershots and replace it? -- Schapel (talk) 15:56, 3 June 2008 (UTC)
 * Opera 9.5 passes more than 10 subtests in each bucket, so I've replaced the example with a Firefox development screenshot instead. -- Schapel (talk) 12:10, 12 June 2008 (UTC)

Mobile Browsers?
Should we add a section regarding the status of mobile browsers in this test? I think some mobile browsers are already doing quite well in this test, for example I think Opera Mobile 9.5 can score over 60/100.

Also now there is an "Acid test" for mobile devices Is your (mobile) browser ready for the Web?, should we have a new entry about mobile browser's standards compliance now? Ufopedia (talk) 07:18, 4 June 2008 (UTC)
 * If you have any information regarding the status of mobile browsers, please contribute and be bold. Ghettoblaster (talk) 10:20, 4 June 2008 (UTC)
 * well, the thing is I can't find reference sources for them, so it will mostly be original research based on my WM6 smartphone. Ufopedia (talk) 11:49, 4 June 2008 (UTC)
 * Uploading screenshots should be okay. Ghettoblaster (talk) 12:08, 4 June 2008 (UTC)
 * I added some mobile browser screenshots Ufopedia (talk) 04:33, 9 June 2008 (UTC)
 * Anyone know if iPhone's 2.0.2 gets a better score than 2.0.1?--HeffeQue (talk) 18:59, 19 August 2008 (UTC)
 * Feel free to create an article on Web Compatibility Test for Mobile Browsers. ff m  16:18, 4 June 2008 (UTC)

Webkit (most of the time and kind of) passes all critera!
The smoothness bug is now closed: https://bugs.webkit.org/show_bug.cgi?id=17510

A couple of bugs remain, all of which are "it fails sometimes or for some users", i.e. reliabality issues.

I am waiting for an official announcement.


 * Yup, I can confirm that the latest WebKit nightly build (r34380) can pass the Acid3 test completely. However there's one interesting thing, according to http://ln.hixie.ch/?start=1207096078&count=1, the performance aspect is supposed to test the overall speed of the browser. Now the WebKit nightly passes all the sub-tests under 33ms, but its overall speed is actually slower than Opera's gogi build. On my C2D@1.6G rig, WebKit takes around 1.8s to pass the test, while Opera takes only around 1.1s, so now we have a strange situation of a browser passing the performance criteria but is overall slower than another browser who doesn't pass. Ufopedia (talk) 11:11, 8 June 2008 (UTC)
 * Should we wait until an official announcement to move WebKit out of the Non-Compliant Section? Or should we wait for a stable release of Safari, which I think will be Safari 4? Because right now, having it under that section and yet stating in the notes that it gets a full pass is rather confusing. -Pyro3d (talk) 18:00, 18 June 2008 (UTC)
 * My idea is that we wait until a stable browser (or other program) based on WebKit is released that passes Acid3. Then we can list the stable versions of applications that pass Acid3. This emphasizes the purpose of the Acid3 test is to have users use browsers that pass Acid3. It doesn't matter what fancy new standards a layout engine provides if there's no stable release to use yet. -- Schapel (talk) 19:51, 18 June 2008 (UTC)

One test is buggy
Test 7 does not follow the spec. https://bugzilla.mozilla.org/show_bug.cgi?id=433662

Either the test will be fixed - an Opera and Webkit will be back at 99, or the spec will be changed. Mozilla probably won't probably implement to the test, but to the spec (as they should).

Until this issue is settled any claim of "passing" the test is preliminary.

--itpastorn (talk) 08:26, 7 June 2008 (UTC)

opera 9.5
Opera 9.5 is out now. It's the official version, however note that they didn't include 100% completion in the released version. This build scores 83%. 81.170.222.15 (talk) 08:33, 12 June 2008 (UTC)

Screenshots from browsers at "time of release"
I'm not sure why this column was added to the non-compliant browsers section. First, it doesn't matter how browsers did when the test was released. All that matters is how they do now. Second, although it's supposedly for showing how browsers did before they started trying to pass specific Acid3 tests, that's not what it shows. For example, the Konqueror screenshot shows how Konqueror did after two months of changes to improve its Acid3 score. Let's remove the column. -- Schapel (talk) 19:46, 8 July 2008 (UTC)
 * Three of the other shots (IE7, FF2, Opera 9.25) show the stable build at the time of Acid3's release, which is what the column should be about. The Acid3 test, at time of release, was meant to have no browsers pass the test, which is what that column illustrates. The Safari screenshot should be showing a stable build, not 3.0 beta, so let's just fix the Konqueror and Safari screenshots, and leave the column intact. nneonneo talk 20:52, 8 July 2008 (UTC)
 * Wouldn't simply stating some scores of browsers (i.e. "At the time of Acid3's release, all browsers performed poorly. For example, Internet Explorer 7 scored 12/100, Safari 3 scored 39/100, and Firefox 2 scored 52/100.") illustrate that point more directly and without crowding the table as much? -- Schapel (talk) 21:12, 8 July 2008 (UTC)


 * It's not just the score that is important, the rendering is also a part of the test. This, I think, is shown quite well by the difference between the screenshot of Firefox and the ones for Opera and Safari, who both have lower "score" but also have a more correct rendering. Also, we can probably remove the "Development Build Screenshot" column once all browsers pass the test. Perhaps the "Notes" column could be moved outside the table, merging it into the lead of the "Layout engines" section for example? I find it to be out of place somehow right now. --Execvator (talk) 15:42, 10 July 2008 (UTC)
 * Actually, the differences in rendering are caused by individual browser choices (when not specified by the test) of font, anti-aliasing, rendering methods, etc. As long as the test result matches the reference rendering as shown in the same browser, the visual test is considered passed. Considering IE's glacial progress, I don't think we'll be expecting it to pass anytime soon, but stranger things have happened.... nneonneo talk 15:46, 10 July 2008 (UTC)


 * As far as I can see, except for text spacing, anti-aliasing and the font-weight for the instructions (and font when Arial isn't available) there really isn't anything left up to the browser to decide. The general rendering should be the same across browsers. --Execvator (talk) 10:32, 11 July 2008 (UTC)


 * Yeah, we can also mention that all browsers rendered the test poorly at the time of release. I think the confusion you're having is thinking that the section is a screenshot section. It is not. It is a section devoted to the current state of layout engines that do not pass. That's why the notes section belongs there (as it gives further details on exactly what does and does not pass), and the screenshots of how browsers rendered the test at the time of release does not. When a stable release of a browser passes the test, the row should be removed, as it would have no place in a section entitled "Non-compliant layout engines". -- Schapel (talk) 16:25, 10 July 2008 (UTC)


 * We already mention that it was written to break in all current browsers at the time. The title of the table is "Layout engine progress for the Acid3 test" which doesn't really exclude those screenshots as they show that the browsers actually have progressed. Though, I'm more interested in seeing one screenshot being preserved as a "how things once were", be it in a history section or elsewhere, so I'm not really against removing that column. Ah, yes, I somehow didn't notice the title. --Execvator (talk) 10:32, 11 July 2008 (UTC)

Deletion of screenshots
Screenshots are deleted or nominated for deletion overzealously. How do I make this stop please? (Rklz2 (talk) 09:25, 25 July 2008 (UTC))

Gecko/2008072703 score is 83 not 84
I get 83 reproducably... (Rklz2 (talk) 09:28, 28 July 2008 (UTC))
 * Are you using Adblock Plus? I get 83 too when I use that add-on. --Spirit55555 (talk) 12:22, 28 July 2008 (UTC)


 * Use safe mode when you take the test or, even better, create a new empty profile specifically for it. --Execvator (talk) 18:06, 28 July 2008 (UTC)


 * No I took the test with a new clean profile... strange because the previous build did score 84 (I uploaded the screenshot), but then the next nightly went back to 83... Just now I tested with a newer one still (Gecko/2008072803) and it's again 84. So anyway, 84 stays :) (Rklz2 (talk) 22:10, 28 July 2008 (UTC))

Opera WinGogi screenshot should not be 99/100
Someone changed the screenshot for Opera's WinGogi build to a 99/100 score - in fact the build scores 100/100 (see the page cited by the text, and many other sources). It indeed fails the smoothness criterion, but the changed image is simply incorrect. I'm reverting it to the original 100/100 image. Luinfana (talk) 17:41, 6 August 2008 (UTC)


 * It should be 99, due to changes to the acid 3 test after the the compatible build was released. If you run WinGogi, it will be 99. The cited page was wriiten before said changes were made. --toehead2001 (talk) 23:41, 6 August 2008 (UTC)


 * (e/c)Actually, the WinGogi build from here now scores 99/100 (verified by myself), with the failure message

Failed 1 of 100 tests. Test 23 passed, but took 46ms (less than 30fps) Test 26 passed, but took 203ms (less than 30fps) Test 46 failed: expected 'uppercase' but got 'none' - case h failed (index 7) Test 69 passed, but took 7 attempts (less than perfect). Total elapsed time: 3.09s
 * I tested this *just today* (and Webkit still scores 100). If you have an Opera build which scores 100, please link it, otherwise, their score stays at 99. nneonneo talk 23:43, 6 August 2008 (UTC)


 * OK, thanks for the correction - I was unaware the test had changed recently. Luinfana (talk) 22:18, 10 August 2008 (UTC)


 * For me, the WinGogi build does get 100/100. *Edit* Actually, it was cached, it does indeed get 99/100. My apologies. El Paulio (talk) 18:36, 16 September 2008 (UTC)

Bot report : Found duplicate references !
In the last revision I edited, I found duplicate named references, i.e. references sharing the same name, but not having the same content. Please check them, as I am not able to fix them automatically :) DumZiBoT (talk) 12:44, 13 August 2008 (UTC)
 * "bug17510" :
 * Bug 17510 - Acid3 test 26 takes >33ms

Screen­shot of Stable Build at Time of Acid3 Release + Safari 3.0 beta??
What's going on here, surely it can't be stable and yet beta.  - ARC Gritt TALK  11:04, 18 August 2008 (UTC)

Nintendo DS Browser
This browser fails with a score of 2/100, like the Opera Mobile browser on which it is based, but it has a significantly different appearance (Which probably means it failed one different test). Unfortunately, I cannot think of any way to get a screenshot, or otherwise prove the score for the sake of Wikipedia. 220.238.82.152 (talk) 03:38, 27 August 2008 (UTC)
 * I think that a picture with a digital camera, resized to fit nicely would be good enough, unless someone knows how to take screenshots off a Nintendo DS.--HeffeQue (talk) 23:18, 28 August 2008 (UTC)
 * Have you tried touching/clicking the "A" of "Acid3"? On Opera Mini, I had to repeatedly click the A until the test updated to the final score, since apparently it would not dynamically update (you can tell it is done if the text says "Total elapsed time:" at the bottom, or if the number of tests failed is equal to the number of points missing. nneonneo talk 02:02, 29 August 2008 (UTC)
 * I've tried it again, and tried your suggestion nneonneo. No change in the score, but touching any part of the word 'Acid3' gives me a partial list of tests. It passes test 5 (in 1000ms), but doesn't tell me about any tests after test 11. The browser is so bad that is can't even display the reference rendering correctly (no boxes are visible).58.105.24.78 (talk) 01:46, 3 September 2008 (UTC)==Google Chrome==

Anyone tried it already on acid3? —Preceding unsigned comment added by 125.60.241.202 (talk) 21:13, 2 September 2008 (UTC)


 * its in the table at the bottom, 78/100--155.144.40.31 (talk) 00:47, 3 September 2008 (UTC)


 * I tested it and got a 77/100, I tought google chrome was a new engine

AbsoluteMSTR (talk) 01:10, 3 September 2008 (UTC)


 * Chrome uses the Webkit rendering engine and a newly developed script engine. So CSS and HTML rendering will be excellent touergh the use of the webkit engine but the sripting moderatly good. hAl (talk) 09:53, 3 September 2008 (UTC)


 * I just got 79 and it says "LINKTEST FAILED" on the bottom. --Mwn3d (talk) 03:08, 3 September 2008 (UTC)


 * Google Chrome on my system (Windows XP SP3) also gets 79 with LINKTEST FAILED...and there is a red box with a picture over the words Acid3 on the page. 20:59, 3 September 2008 (UTC) —Preceding unsigned comment added by Congdonweb (talk • contribs)

Should Chrome be listed in the same category as Safari? Even though they might use the same rendering engine, it does score differently than Safari... —Preceding unsigned comment added by Congdonweb (talk • contribs) 20:08, 3 September 2008 (UTC)


 * That row is for WebKit not Safari.. We will never have a pic of Chrome for the 1st column, and WebKit already passes so the 3rd column will be the same, so maybe when Chrome is final it could replace the Safari screenshot in the 2nd column, until then I don't see any point in adding it, because it uses an older version WebKit.. ....Alternatively the whole table could be made browser-oriented instead of browser-engine oriented. (Rklz2 (talk) 20:10, 4 September 2008 (UTC))


 * Making rows for rendering engines alone is now showing to be useless. ACID 3 is not just about the rendering engine but also about the scipt engine. Adding Chrome to the webkit row is incorrect as it uses a different script engine and the result might not be the same. hAl (talk) 21:20, 4 September 2008 (UTC)


 * As far as I know Chrome still uses KJS aka JavaScriptCore aka WebKit's JavaScript engine, it just has a JIT built into it like Tamarin nanoJIT into SpiderMonkey in the upcming Firefox 3.1, but functionally this should not change anything, only speed up commonly executed code. Maybe this is wrong... The score is different because Chrome uses a different version of WebKit than Safari currently does. (Rklz2 (talk) 02:24, 5 September 2008 (UTC))
 * Hm, have you heard of V8? I wonder what your sources are. --Kjoonlee 04:29, 5 September 2008 (UTC)


 * Isn't V8 just JavaScriptCore with a JIT added to it? I thought it's not a totally new engine. (Rklz2 (talk) 17:58, 5 September 2008 (UTC))
 * Why do you think so? --Kjoonlee 23:53, 5 September 2008 (UTC)


 * Common sense. The time in which Google developed V8 is way too short to develop a whole new engine. And there are perfectly good ones already; They chose WebKit instead of building a new HTML renderer too. Unfortunately I can't find Chrome's source anywhere despite Google saying it's open-source: the source directory on their Google Code site is just empty. Anyway, I think testing the same, older version of WebKit, using a nightly build, that Chrome uses should give identical results to Chrome, if I'm right... (Rklz2 (talk) 04:27, 7 September 2008 (UTC))
 * No sources, then. You can check out http://v8.googlecode.com/svn/trunk/ and you can see that besides regexp related code, there's nothing which indicates JavaScriptCore. --Kjoonlee 05:19, 7 September 2008 (UTC)


 * I have found Google Chrome's (rather well-hidden) source code. You were right, anyways since then someone added the table in the article a separate row for "WebKit with V8". (Rklz2 (talk) 04:31, 12 September 2008 (UTC))

Formal Acid 2 Test page
The formal Acid 2 Test page is http://www.webstandards.org/files/acid2/test.html, not the other one. The other ones causes to fail the test. For more info read this: http://blogs.msdn.com/ie/archive/2008/03/05/why-isn-t-ie8-passing-acid2.aspx --144.122.250.219 (talk) 13:07, 23 September 2008 (UTC)


 * This bug was resolved in Internet Explorer 8 beta 2. —Remember the dot (talk) 14:39, 23 September 2008 (UTC)

Chromium build scores 100/100 today
the merge with webkit@36102 in build 2778 brings it up to 100/100

http://build.chromium.org/buildbot/snapshots/chromium-rel-xp/2778/changelog.xml

Ufopedia (talk) 14:41, 2 October 2008 (UTC)
 * I think this demonstrates that Chrome should not be considered different (for this article) from Safari/WebKit. The JS engine has very little to do with it. nneonneo talk 15:09, 2 October 2008 (UTC)


 * Indeed... It no longer makes much sense for them to use "V8" anyway as WebKit has it's own JIT now... (Rklz2 (talk) 22:50, 2 October 2008 (UTC))


 * Well, I think there are probably reasons beyond just the JIT to keep V8 for Chrome; for example, V8 is still faster on pure-JS benchmarks (though Safari's nightlies are far better at DOM+JS stuff than Chromium nightlies). Anyway, I do recommend that Chrome be kept off this page, unless they do something drastic like using a custom rendering engine (which, of course, is rather unlikely). nneonneo talk 03:31, 4 October 2008 (UTC)


 * It looks like they are moving toward a JavaScriptCore-based build of Chrome with r2863, especially this change (Rklz2 (talk) 19:23, 4 October 2008 (UTC))


 * Huh. That is actually quite surprising. Also, I tested a recent WebKit nightly, and found it ran only around 25% slower on Chrome's own V8 test (http://code.google.com/apis/v8/run.html). This is quite rsurprising, considering that when Chrome was released, WebKit nightlies were only scoring a third of what Chrome was achieving. Maybe JSC's speed gain has convinced Google to use it instead. nneonneo talk 22:04, 4 October 2008 (UTC)

NetSurf
Should we really include NetSurf? Their own page states that JavaScript support is still a long ways off, and until they support JS, they don't have a hope of even starting the test. It would be appropriate to put this browser up for Acid2, which does not feature JS, but for Acid3 I don't think it is appropriate. nneonneo talk 22:01, 4 October 2008 (UTC)


 * That's true, so I will comment it off... (Rklz2 (talk) 23:04, 4 October 2008 (UTC))