Talk:Acid3

Archive

Old discussions for this page has been archived.

• Talk:Acid3/Archive 1 —Preceding unsigned comment added by Itpastorn (talk • contribs) 18:34, 8 April 2008 (UTC)
• Talk:Acid3/Archive 2 mabdul 0=* 18:50, 7 December 2008 (UTC)
• Talk:Acid3/Archive 3 - ARC Gritt^TALK 20:50, 27 November 2009 (UTC)

Chromium fails the acid3-test

As of July 3 2009, Chromium now passes Acid3 100%. Any build of version 3.0.192.0 (build 19910) and above now passes. You must load chrome with the ' --enable-remote-fonts flag in order to get a pixel-perfect rendering (remote fonts are disabled by default in Chrome due to security reasons according to http://code.google.com/p/chromium/issues/detail?id=231#c85 —Preceding unsigned comment added by 99.188.54.154 (talk) 04:51, 4 July 2009 (UTC)

Chromium says 100/100 but the result I get is not the same as the reference:
- Chromium shows incorrectly "LINKTEST FAILED".
- The grey shadow of "Acid3" is missing.
213.10.195.214 (talk) 20:24, 4 December 2008 (UTC)

Be sure you are using the same build of Chromium referenced in the article. Luinfana (talk) 20:19, 8 December 2008 (UTC)

My results for that build are the same as for the latest build. There really seems to be a problem with the linktest. Chromium also fails another test: http://www.css3.info/selectors-test/test.html It probably has something to do with: https://bugs.webkit.org/show_bug.cgi?id=22131 213.10.195.214 (talk) 14:06, 9 December 2008 (UTC)

As of build 7750 chrome STILL doesn't pass all parts of the acid 3 test (they fixed font but linktest failed is still there) so this should definitely be fixed. Zamadatix (talk) 20:38, 8 January 2009 (UTC)

I get 99/100 with Chromium 2.0.180.0 when I go to http://acid3.acidtests.org/ from the start page showing most visited pages. But then I get 100/100 when I hit reload. Termopetteri (talk) 11:32, 19 May 2009 (UTC)

Instructions for taking the test are to visit the page and press reload. The first result (before reload) is expected to be often incorrect. _ɹəəpıɔnı 00:16, 20 May 2009 (UTC)

I just ran it with 4.0.249.4 on Ubuntu 9.10, and the rendering is pixel-perfect, but not entirely smooth. --86.69.114.171 (talk) 18:21, 23 November 2009 (UTC)

Safari 4 does not pass Acid3

In the same way that Opera 10.0 fails the performance test, Safari 4 does too. Even worse, it takes over double the time on the same harware (Core2 Duo E6600). Linky. So by the same token the iSheep are up in arms that Opera should be removed, so should Safari. —Preceding unsigned comment added by 81.174.171.21 (talk) 11:36, 2 September 2009 (UTC)

We cannot say that Safari does not pass in the article based on original research. You need to find a reliable source to cite. Also, please remember to assume good faith. Thank you. -- Schapel (talk) 12:17, 2 September 2009 (UTC)

Whether or not a browser passes Acid 3 depends on whether somebody makes an official statement? Sounds a bit weird to me. BTW, I also never could find out how Safari 4 "passes" Acid3 better than Opera 10 does. But then, this discussion looks so ridiculous to me that I don't really want to get involved. --DeTru711 (talk) 17:17, 3 September 2009 (UTC)

Whether or not we can say it in Wikipedia depends on whether somebody makes an official statement, because Wikipedia requires a citation from a reliable source and does not allow original research. -- Schapel (talk) 17:52, 3 September 2009 (UTC)

So if somebody makes an official statement that Safari 4 passes Acid 3 qualifies it to put it on Wikipedia, even if that statement is "wrong" (since Safari 4 doesn't really seem to pass the performance aspect of Acid 3 in the same way that it's demanded for Opera)? To me it looks just like a battle between fans of either browser, with Safari fans wanting their browser to be the only passing one (BTW, I'm not a fan of either; only the reasoning behind the decision looks ridiculous to me because obvious facts that everybody can test for himself should count more than official statements by somebody). Bah, I didn't want to engage in this discussion, now I'm afraid I did :(. But be relaxed, I'm not going to change anything on the page regarding this question - I leave that up to you guys to battle that out. --DeTru711 (talk) 18:27, 3 September 2009 (UTC)

Sources are supposed to be third party, the citation stating that safari passes is currently from the apple website. They could easily have ignored the performance aspect, the never specifically state what the performance is. Maybe we should ignore the performance aspect in judging whether it passes due to the difficulty in testing. Acid3 is really meant to test standards compliance, not speed, anyway.-AlexTG (talk) 00:44, 4 September 2009 (UTC)

This sounds very reasonable to me. All the test says is "the animation has to be smooth". If that applies to one of them, it certainly applies to the other as well (or conversely, to neither). --DeTru711 (talk) 01:30, 4 September 2009 (UTC)

Perhaps the best way to handle this is to change the section header to something like "Browsers which score 100%", and then list all the browsers that do. Next to Safari, we can add an extra note that Safari is known to also pass the performance measure as well. I think this seems much more reasonable, especially given the discrepancy in the evidence. Does this seem like a change that would make everyone happy? (I, personally, would feel much better) Ean5533 ( View! / Talk!) 13:06, 4 September 2009 (UTC)

It sounds fine to me, except that I would add passing the rendering aspect of the test to the criterion. We could then have a note explaining which browsers pass the performance aspect of the test, and even state how "close" a browser is if it doesn't pass, provided that a citation to a reliable source can be given. -- Schapel (talk) 14:16, 4 September 2009 (UTC)

Why not say that the performance aspect is disputed (instead of any Yes or No), with a side note saying that this is not accurately defined which leaves open several ways of interpretation? As this discussion proves, that would describe the facts best. --DeTru711 (talk) 17:00, 4 September 2009 (UTC)

Ian Hickson, the author of the Acid 3 test, very clearly states that the performance aspect of the test is based on using "... whatever the top-of-the-line Apple laptop is at the time the test is run" and not getting any error messages such as "test XX took XXms (less than 30 fps)." While I personally think that metric is complete garbage and totally biased/unfair, it is the test author's choice. If we want to place value in his test, we have to follow all of his rules.

And that's fine, but I'm pretty sure you can throw a x64 32 CPU mega-server with Opera 10 (being a top-of-the-line server), and it will also pass the Performance test. Referring back to my original link, Safari is consistently taking 1-2 seconds more to pass the Acid3 on typical hardware, and it's rather misleading to suggest that Safari passes the performance aspect and Opera does not. Setting them both to NO and citing my original benchmark screenshot. Awaiting the minutes before the iSheep put it back... —Preceding unsigned comment added by Mgillespie (talk • contribs) 22:31, 20 November 2009 (UTC)

I reverted your edit because one needs to run the test on the reference platform (a top of the line Apple laptop), and because original research is not allowed in Wikipedia. Also, please assume good faith. Thanks. -- Schapel (talk) 22:48, 20 November 2009 (UTC)

However, I do think it's fair to mention it in the article. Maybe in the section where we list browsers that pass, we can make a note about the performance metric using a top-of-the-line apple laptop as the reference hardware. That will make it obvious to anyone that cares why Opera or other browsers are potentially handicapped. Ean5533 ( View! / Talk!) 18:11, 4 September 2009 (UTC)

Epiphany

151.59.216.108 is right. Epiphany 2.28 released on 1 November 2009, part of GNOME 2.28, pass the test with 100/100. This happen also because of the switch in rendering engine from Gecko to WebKitGTK+. To find a reliable source I simply followed the link in Epiphany Wikipedia article:
http://arstechnica.com/open-source/news/2009/09/linux-garden-gets-a-new-gnome-with-version-228.ars
http://arstechnica.com/open-source/news/2008/04/webkit-gtk-port-passes-acid3.ars
Then I just tried to compile and install Epiphany by myself. Try yourself. As today is the only WebKit solution for Linux that pass Acid3 test --Efa (talk) 23:16, 9 November 2009 (UTC)

The only information I can find about how Epiphany does on the Acid3 test is that it currently scores 100/100. There is no information about rendering or performance, which are the other two aspects of the test. -- Schapel (talk) 14:05, 10 November 2009 (UTC)

To be true, the links speak about the fix done to get perfect pixel rendering past in 2008. Anyway for the performance part test, you are right, I have no realiable source. I will write to ars technica asking for a clarification.

About that I'm in favor of rename the paragraph to "Browsers which score 100" (and so add Google Chrome and Midori), add a column "Pass" and write Yes when all the 3 criteria (Score 100/100, Pixel-perfect rendering and Performance) are all Yes from a reliable source. This clarify the confusion about the 100% of many users.--Efa (talk) 23:26, 11 November 2009 (UTC)

The rendering under Ubuntu 9.10 is not Pixel-perfect, although it's damn close. Looks mostly like differences in line spacings. --86.69.114.171 (talk) 18:17, 23 November 2009 (UTC)

Why the anti-Opera bias?

I thought there was supposed to be NPOV on Wikipedia. The Safari passed the performance tests and Opera does not is totally bogus, and seems to get constantly reverted back by some kind of Apple fanboi. Both companies claim 100/100 Acid3 compliance, neither elaborate as to if that includes the performance tests, yet it seems in the case of Apple you can infer it does, in the case of Opera it does not, DESPITE evidence suggesting that Safari performs significantly worse than Opera on the same hardware...

I know Opera has a low marketshare in the US, and Apple is an American company, and Americans like to support homegrown stuff, but showing extreme bias like this under compelling evidence that suggests it' bogus is just embarrassing (for the author that seems to want to keep reverting the edit). —Preceding unsigned comment added by 195.75.83.25 (talk) 09:33, 23 November 2009 (UTC)

If you can find a reliable source that states that Opera passes the performance aspect of Acid3 (a reliable source stating that Opera passes Acid3 would imply that Opera passes all aspects), feel free to update the article. No one has been able to find one yet, only sources that say that Opera scores 100/100 and has renders the test correctly. There is no bias -- we are just following Wikipedia's guidelines, such as verifiability and citing reliable sources. As for yourself, please do the same and assume good faith. Thanks. -- Schapel (talk) 14:39, 23 November 2009 (UTC)

It's because the safari citation has an apple representative saying "safari passed acid 3" while the opera citation says "100% score on Acid3 test". Hence assuming the apple representative knew about the performance aspect and was providing the quote in good faith, safari must have passed the performance aspect. I think it's likely he just wasn't aware/didn't care about the performance aspect but meh -AlexTG (talk) 03:21, 24 November 2009 (UTC)

as per WP:Third-party sources none of the citation are reliable. I'm not questioning apple's claim but changing both Safari and Opera to show they passed based on their claim and adding the citation needed template to both. Chris Ssk _talk 20:26, 28 November 2009 (UTC)

The guideline you refer to does not say that first-party sources are not reliable sources, but that you cannot rely solely on first-party sources when you write an article. In other words, articles need citations from third-party sources. This article has plenty, so it meets the guideline. -- Schapel (talk) 01:15, 29 November 2009 (UTC)

If you have a 3rd party source then put a note next to the "Yes" to show it, and if you have a source that says that Opera fails the performance aspect of Acid3, change it to "No" and again put the link. Both entries rely on press releases as sources, as per the guideline I posted above, such sources should be identified as claims, not facts. It makes no sense to me to take Apple's claim as fact but reject Opera's claim. As it stands both companies claim their browser passes the test, no 3rd party sources are available to show otherwise. Both should be set at Yes with {{citation needed}}. Chris Ssk _talk 07:33, 29 November 2009 (UTC)

No, the guideline you referred to above has nothing to do with the edits you're making. You can see the discussion above for all the details. The bottom line is that the source says Safari passes Acid3 (meaning it passes all three aspects), but the source for Opera only says it scores 100/100 (it passes only one aspect). Please discuss the issue before making further changes. Don't just make the change and post a message. Thank you. -- Schapel (talk) 13:48, 29 November 2009 (UTC)

How are the guidelines not relevant? Wikipedia:Third-party_sources#Non-independent_sources says: "Non-independent sources may be used to source content from articles, but the connection of the source to the topic must be clearly identified."..."Any publication put out by an organization is clearly not independent of any topic that organization has an interest in promoting." yet you insist on removing the {{citation needed}} from the Safari entry saying that Apple's website is sufficient source. Furthermore while you accept apple's claim as fact you reject opera's claim. Chris Ssk _talk 18:45, 29 November 2009 (UTC)

If you want to follow the guidelines, then say that Apple states that Safari passes the test. Don't say that there's a citation needed when a citation has already been provided. Once again, I do not reject Opera's claim. I fully acknowledge that Opera gets a score of 100/100 on the Acid3 test. What is in question, however, is whether Opera passes the performance aspect of the test. The latest source we have says that Opera does not. -- Schapel (talk) 20:42, 29 November 2009 (UTC)

Yes, Apple states that Safari passes the test, and there is a citations for that. where does it say it has no timing issues? As noted on the Acid3 creator's blog [1] "The idea is to make sure that browsers focus on performance as well as standards. Performance isn't a standards-compliance issue, but it is something that affects all Web authors and users. If a browser passes all 100/100 subtests and gets the rendering pixel-for-pixel correct (including the favicon!), then it has passed the standards-compliance parts of the Acid3 test. The rest is just a competition for who can be the fastest." So Safari passing the test does not equal not having timing issues and Apple saying that they pass the test and Opera saying they got 100% (ie got 100/100 and rendered pixel-for-pixel correct) is the same thing. In fact I think both entires are wrong and should be changed to N/A or the performance column should be removed all together. Chris Ssk _talk 11:41, 30 November 2009 (UTC)

In saying that Safari passes the Acid3 test, that means there are no timing issues. -- Schapel (talk) 19:31, 1 December 2009 (UTC)

did you even read the quote I provided above? Chris Ssk _talk 20:34, 1 December 2009 (UTC)

I don't think this table is working out. The fact is that the performance test is very ambiguous and completly dependant on the computer it is run on. So really, unless we get a reliable source which compares the browsers on many different computers we can't be sure of anything. Right now we have two first party sources, neither of which specifically refer to the performance test and both of which would have been run on different computers. I would prefer we removed the table, listed the two browsers that passed and had a note stating that the performance aspect is ignored due to ambiguity of the test.-AlexTG (talk) 21:37, 3 December 2009 (UTC)

Its quite clear how acid3 performance is tested. As mentioned in the article the "reference hardware" as set by the test creator is whatever the top-of-the-line Apple laptop is at the time the test is run.[2]

However, I agree that the table is not working, it confused me. if its for browsers that pass the Acid3 test (Note:For a browser to pass the acid3 test, it must have a 100/100 score, the final page must be identical to the reference rendering, and the animation has to be smooth (all tests less than 33ms)), then only Safari should be listed because as far as we know Safari is the only one to claim to pass all 3[3].

I think the "Browsers that pass" section could be changed to "Compliant Browsers" with a note explaining the performance aspect. Both Safari and Opera have passed the standards-compliance parts of the test (100/100 and rendering). --Chris Ssk _talk 12:22, 4 December 2009 (UTC)

I think that the reliable source that say "browser X pass the test", should pubblish also a shot of the dialog box that appear clicking on A, that show the perfect timing on the reference hardware. Without that shot, we cannot really believe that they mean pass all the three parts of the test and not only the 100% compliance.--95.74.56.172 (talk) 21:50, 8 December 2009 (UTC)

The Wiki article should show this shot for passing browser too.--Efa (talk) 23:47, 10 December 2009 (UTC)

Here's a thought: why not say before the table that the sources come from first-party documents? That way, everything fits Wikipedia's rules about first-party sources without any subjectiveness. --Unknownwarrior33 (talk) 04:28, 15 December 2009 (UTC)

Updating

What's the standard for updating results? I got a better result than listed for a browser, but I fear that if I post a new picture, it will be taken down as original research. So what have others been doing? --Unknownwarrior33 (talk) 16:19, 23 November 2009 (UTC)

I think others have been posting screenshots from their browsers. I have been checking to see that the screenshots agree with what others have said and match screenshots from browsertests. If I have no reason to believe that a screenshot does not accurately portray how a browser renders Acid3, I let it stand. I would avoid interpreting the results of the test, such as claiming that the rendering is correct or stating that the browser passes the performance aspect, as that crosses the line to original research by synthesizing data from multiple sources. -- Schapel (talk) 17:37, 23 November 2009 (UTC)

Screenshot At Time Of Release

Why are the now-compliant browsers not listed with their "Screenshot at time of release"? These should be included for posterity. - ARC Gritt^TALK 20:44, 27 November 2009 (UTC)

This has annoyed me since that change was made, and I agree. Fix it.--82.182.69.67 (talk) 22:25, 4 December 2009 (UTC)

Internet Explorer 8 results?

I'd just like to chime in and say that it's silly that Original Research is not considered in this article. The Internet Explorer 8(on Windows 7) results I got completely disagree with this article's opinion. First off, I had to install a Microsoft plugin to even properly test the browser (which I think completely nullifies any results on this browser). After installing the plugin I got 16/100, then 20/100, then 6/100. The 6/100 was completely consistent, even after restarting my computer. I use IE8 as my main browser(I know...firefox bla bla) and I think that any MSIE8 figures should be removed or edited. Hubertforthcourse (talk) 08:13, 3 December 2009 (UTC)

The core content policies of Wikipedia are No original research ,Neutral point of view and Verifiability and cannot be superseded by other policies or guidelines, or by editors' consensus. No offense but how can we verify that you ran the test with the correct settings? (On a site note, when I run the test in IE8 I get 20/100 in XP, Vista and 7). Also you cant remove facts from an article because you don't agree with them --Chris Ssk _talk 13:12, 3 December 2009 (UTC)

Where is the bar?

I tried to remove Konqueror from this list, but my edit was undone, so I'd like to get a discussion started here. Konqueror has .03% of the market share. I know .03% is just a number, and there's no sense in picking a random number and using that as the bar. However, that is less than a huge number of other browsers not on this list (i.e. Netscape, and "unknown" beat out Konqueror, 70:1 and 20:1). Also, it's unsourced which means the burden of evidence lies with users wanting to add/keep the material. I also think even mentioning it violates undue weight: "generally, the views of tiny minorities should not be included at all". ~a (user • talk • contribs) 15:28, 7 December 2009 (UTC)

Market share is not the factor to determine the importance of the browsers listed. Acid3 tests are about the DOM and JavaScript, so basically the test is about layout engines and JavaScript engines. The main layout engines for web browsers are: Trident used by IE (JScript), Gecko used by Firefox (Spidermonkey JavaScript engine), WebKit used by Safari and Chrome (Safari uses JavaScriptCore as its JavaScript engine while Chrome uses its own V8 JavaScript engine), Presto used by Opera (Futhark JavaScript engine) and KHTML used by Konqueror (KJS JavaScript engine). There is no point in listing Netscape or Flock or Avant or Epiphany etc... Any Gecko-based browser will score the same as the Firefox version it is based on, same with WebKit-based browsers, Internet Explorer shells will score the same as IE. Konqueror uses a layout engine that is not represented by by any of the other browsers listed. --Chris Ssk _talk 22:04, 7 December 2009 (UTC)

Well I agree with most of what you said but you didn't reply to any of my arguments. This still fails V and UNDUE, right? My market share argument (though I agree it shouldn't be arbitrary) was because UNDUE specifically states "generally, the views of tiny minorities should not be included at all". ~a (user • talk • contribs) 00:14, 8 December 2009 (UTC)

Views aren't being expressed in this situation, facts are. We are comparing the different performances of different browsers/engines, rather than providing differing opinions of the browsers.-AlexTG (talk) 02:16, 8 December 2009 (UTC)

Alex, using your logic, do we need to list all of the browser/user agents? Even the ones that almost nobody uses? Do we need to list NetSurf? Do we need to list Tkhtml? AWeb? The Bat? Entorage? None of them use layout engines discussed above. Also no one seems to be discussing verifiability. It was not sourced, I removed it (citing among other things verifiability) and somebody added it back in. I'm pretty sure that's not allowed. ~a (user • talk • contribs) 15:36, 8 December 2009 (UTC)

Acid3 is a web browser test, I don't think it applies to e-mail clients like the The Bat or Entourage. If you think that a browser should be on the list but its not then please add it, Its my understanding that the 5 engines listed in my comment above (along with Tasman that is no longer used in a web browser) are generally accepted as the notable layout engines, and each of these engines is represented by its lead browser. Also if its an issue of verifiability of a claim then just add {{citation needed}} to it, don't remove the entry, and why only delete Konqueror? its not like the reference links for the rest are better. --Chris Ssk _talk 21:32, 8 December 2009 (UTC)

"Its my understanding that the 5 engines listed ... are generally accepted as the notable layout engines" Ok. I guess I'm willing to concede that point since it's starting to look like I might be in the minority here. However, your second point that "its not like the reference links for the rest are better". That argument I still would like to disagree with. Any content that fails V can be challenged/removed by any editor regardless of how the other content stands up to V. ~a (user • talk • contribs) 22:38, 8 December 2009 (UTC)

You are of course right about Verifiability, I was just making an observation that not only Konqueror but almost all the browser results are unreferenced. Here's a quick overview of the issues they have.

• Chrome has no reference links
• Firefox and SeaMonkey are unreferenced for the "release at time of Acid3 release" and "current release", the "preview release" has a screenshot and score of 3.7a1pre Minefield but the reference link is for 3.6a1pre Minefield.
• Konqueror is also unreferenced for the "release at time of Acid3 release" and "current release", current and preview release info is outdated, and the reference link of the preview release is for a version older than the stable release, Screenshots are also for older releases.
• IE is unreferenced for the “release at time of Acid3 release” and “current release”

Mobile browsers seem to also have the same problems. --Chris Ssk _talk 19:53, 10 December 2009 (UTC)

Chrome (Not Chromium)

For the latest Dev build of Chrome (Google Chrome 4.0.266.0), everything is perfect. The picture shown in the page shows that. Why is it listed as failing?

--Unknownwarrior33 (talk) 23:39, 13 December 2009 (UTC)

Whoops, nevermind; just re-read the page. --Unknownwarrior33 (talk) 23:45, 13 December 2009 (UTC)

Fixing disagreements

I understand completely why there is so much disagreement on this page. I'm not sure if this suggestion fits with Wikipedia's policies, but might it be useful to note the uncertain nature of these results on the page, near the chart? It would, if nothing else, make things clearer. The performance aspect, and the first-party statement from Apple, mean the listed results are not 100% perfect and exact, so people reading the page should be aware of that. --Unknownwarrior33 (talk) 23:49, 13 December 2009 (UTC)

Opera Mobile 10 "beta" status

Firstly, Opera Mobile 10 is not listed in any of the tables in this article. I don't know what it scores, but it has the newest Presto 2.4 engine (the latest desktop browser contains Presto 2.2) which makes it quite notable I think.

However, this presents another issue. The latest "stable" version of Opera Mobile listed is 8.65. However, Mobile 9.7x has never been released as "final", and 10 is now out. Also, despite the current release being advertised as "Opera Mobile 10 beta 2" on the Opera Software website, it is the ONLY version officially available for download via opera.com. This makes me wonder if Opera are now using some "perpetual beta" type of versioning system for the Mobile product, though they have made no statement to this effect.

Which version should be included as the "current release" and "latest preview release" respectively in the table? Opera Mobile 10 beta 2 is definitely the "current release" if we're to go by what is the main official release provided for download. _ɹəəpıɔnı 03:09, 14 December 2009 (UTC)

As far as I know Opera Mobile 10 is still only available in beta, however while 8.65 was the latest stable release that was available to download newer versions have been available pre-installed on many phones. It seems its Opera's strategy to release public betas but then only sell the finished product to manufacturers (sales of Opera Mobile is one of their major sources of income). HD2 ships with Opera Mobile 9.7 therefore I think Opera Mobile 9.7 is RTM and so a released product, with a score of 100/100, correct rendering and performance not an issue Opera Mobile could be moved to the "Browsers that pass" section --Chris Ssk _talk 09:39, 14 December 2009 (UTC)

Firefox

Firefox only passed with a score of 93 NOT 96. I can take a screenshot if you want me to. --Tyw7  (Talk • Contributions) 10:56, 14 December 2009 (UTC)