The Acid3 test has a rendering subtest that checks the positioning of text in particular conditions (absolutely-positioned generated content with embedded fonts). To get precise results, I used a single glyph from the Ahem font, which has well-known metrics. My plan was to have the glyph set up so that a perfect white 20×20 pixel square glyph from the font would be overlaid exactly on a 20×20 pixel square red background, thus hiding everything when things lined up. I positioned this test in the upper right hand corner, snug against the black border of the test.
The problem with this test is that on some platforms, specifically Mac platforms with LCD antialiasing, the font rendering system actually renders the glyph using sub-pixel effects, which ends up overlapping the border and makes the test not look the same as the reference rendering.
This would affect any browser, but only on Mac. Unfortunately, the WebKit team "fixed" this problem by simply hard-coding the Ahem font and making it not antialias.
Now, I argue that this is a bug in the antialiasing, but sadly there's no real spec for the antialiasing and so other people argue that it shouldn't be in the test in the first place, whether I'm right or wrong. What we all agree on is that the font-specific hack is lame. (It's especially bad with this font because Ahem is supposed to be a testing font and we specifically don't want it going down different codepaths!)
So Hyatt and I came to a deal. I would move the test down and to the left one pixel, so it doesn't affect the border anymore, he would accept to remove the hack, and would fix one additional bug (a background-position rounding bug).
This doesn't affect Opera's news of their progress, as their build is currently only available on Windows.
The test will probably have a few more minor changes as people track down the last few remaining problems, in particular in the SVG subtests and on the performance part of test 26. (Test 26 is supposed to track the incremental speed of computer hardware, but it hasn't been really calibrated well yet, I just estimated what the numbers should be.)
It's great to see WebKit and Opera work so hard on interoperability issues such as those brought up by Acid3. The Microsoft and Mozilla teams are currently in the "crunch time" of their respective browsers' releases, so it's expected that they wouldn't be working on this at this time — at the end of a release cycle, stability, performance, and user experience are usually much more critical. Hopefully once IE8 and Firefox3 are released they will be able to turn once more to the world of standards and we'll see big improvements of the Web platform again.
Pingbacks:
1
Since we announced the Acid3 test a few short weeks ago, two major browser vendors have been tripping over themselves trying to pass the test, with both Opera and Safari getting very close in the last few hours.
Of course, with a test as complex as this one, I was bound to make some mistakes, which the browser vendors were very quick to tell me about. Sometimes they were mistaken, and the test was actually correct; other times I was wrong, and had to fix the test, and in one case, the spec changed.
Here's a quick overview of the more major changes I made to the test. Luckily, none of the errors were too serious.
- Sub-pixel testing
- It turns out that the original test accidentally required that browsers implement sub-pixel positioning and layout (and in fact the reference rendering got it wrong too, and relied on the same kind of rounding as Firefox does), which is somewhat dubious. I've changed the test to not rely on sub-pixel layout. However, it is very likely that this will be tested in Acid4, if we can get the specs to be clearer on this.
- Surrogate pairs in SVG APIs
- One of the submitted tests assumed that SVG APIs worked on Unicode codepoints, but the SVG spec changed to work on UTF-16 codepoints, like the rest of the DOM API, so the test was changed there. (The test changed a couple of times, because I originally got the fix wrong.)
- The
click() method
- The test originally assumed that the
click() method was reentrant, but the specs were vague on this and someone suggested making it fail if calls to it were nested, so I removed this part of the test (the spec hasn't been updated yet). I replaced it with an attribute test (the new second part of subtest 64).
- The Performance Test
- I made the loop counter in the performance test (a part of subtest 63) less complicated and shorter, to make it at least plausible that browsers could be fixed to pass that test quickly enough that it wouldn't always feel jerky. At the same time, I updated the test's infrastructure to report more details about pass and fail conditions and how long each subtest takes to run.
- Namespace bug
- Someone noticed that
http://www.w3.org/1998/XML/namespace should have been http://www.w3.org/XML/1998/namespace in one of the subtests.
- Linktest timeout
- I made the linktest more resilient to slow network conditions. However, the test is still going to give you major issues if you are on a network with multi-second latency, or if the acidtests.org site is being slow.
When we released Acid2, the first browser passed it in about a fortnight. Acid3 is orders of magnitude more complicated. I really didn't expect to see passing browsers this side of August, let alone within a month. I am really impressed by the level of commitment to standards that Opera and the WebKit team (and Apple in particular) are showing here.
Pingbacks:
1
2
3
4
5
6
After baking it for several weeks, we have finally decided Acid3 is stable enough to announce
it is ready. We'll be working on a guide and other commentary in
the coming weeks and months, but it'll take a while — Acid3 is
far more complex than Acid2 was.
I have to say straight up that I've been really impressed with the
WebKit team. Even before the test was finished, they were actively
following up every single bug the test showed. Safari 3 (the last
released version) scores 39/100 with bad rendering errors. At the
start of last month their nightly builds were scoring about 60/100
with serious rendering errors. Now barely a month later the nightly
builds are already up to 87/100 and most of the rendering errors are
fixed. That's a serious testament to their commitment to
standards. (Also, I have to say, it was quite difficult to find
standards compliance bugs in WebKit to use in the test. I had to go
the extra mile to get WebKit to score low! This was not the case with
most of the other browsers.)
Speaking of standards, and of good news from browser development
teams, Microsoft's IE team announced
today that they were changing their mind about their mode switch, and
that bug fixes they make to their rendering engine will be
applied to pages in what HTML5 calls "no-quirks" mode (what has
historically been known as "standards mode"). I'd like to congratulate
the IE team on this brave decision. It's the right thing for the
Web.
Meanwhile, HTML5
continues to make good progress. I have a page now which shows my progress in replying
to e-mails, and as you can see from the changelog
tracker, checkins are as fast and furious as ever.
Pingbacks:
1
2
3
4
5
6
7
8
There were several announcements yesterday.
First: The W3C finally got around to publishing one of the drafts
of the
HTML5 spec as a "first public working draft" (a misnomer if ever
there was one, given the history of HTML5). This is news for one
reason and one reason only: it starts the W3C patent policy clock,
which means that within a few months we will know if any of the
browser vendors have patents they wish to use to block progress on the
Web. Still, a step forwards.
Second: The syntax for Microsoft's plans for yet another
quirks mode switch in IE8 were announced. Basically they are
offering authors the option to pick a specific set of bugs, defaulting
to IE7's. The idea is that with each new major IE version, they can
decide to simply freeze their last set of bugs forever.
If Web authors actually use this feature, and if IE doesn't keep
losing market share, then eventually this will cause serious problems
for IE's competitors — instead of just having to contend with
reverse-engineering IE's quirks mode and making the specs compatible
with IE's standards mode, the other browser vendors are going to have
to reverse engineer every major IE browser version, and end up
implementing these same bug modes themselves. It might actually be
quite an effective way of dramatically increasing the costs of
entering or competing in the browser market. (This is what we call
"anti-competitive", or "evil".)
It will also increase the complexity of authoring by an order of
magnitude. Big sites will become locked in to particular IE version
numbers, unable to upgrade their content for fear of it
breaking. Imagine in 18 years — only twice the current lifetime
of the Web! — designers will not have to learn just HTML,
they'll have to learn 4, 5, maybe 10 different versions of HTML, DOM,
CSS, and JS, just to be able to maintain the various different pages
that people have written, as they move from job to job.
I mentioned
all this on public-html last April, pointing
out that if Microsoft was really interested in not breaking the
Web they would instead send technical comments on the HTML5 spec,
explaining what exactly the spec requires that is incompatible with
deployed content. (Microsoft have yet to send any technical feedback
on HTML5 at all, 8 months later.)
I've been shocked to see several people in the industry supporting
this move, including people whom I have met personally and whom I know
know better, but I'm in no position to criticise people who should
have known better. I'm glad to see that at least people from Opera, from
Mozilla (continued),
from
Webkit, and even from the
WaSP, and from the community as a whole (if the comments on
Slashdot and other sites are representative) see this for what it
is.
There are several directions we can go in from here. We could
prematurely claim that our pages want IE8's new quirks mode, which
would likely cause pages to change rendering in IE8 as compared to
IE7. However, that's mostly a futile exercise, as either few people
will do this, and it won't matter, or a lot of people will do this,
and Microsoft will just change the syntax before IE8 comes out.
We could use the "edge" feature that is apparently being provided
to always trigger the latest mode. However, that would be even worse:
if enough people use this feature, then Microsoft will feel compelled
to just make "edge" equivalent to "IE8" and introduce a new way to
trigger the "latest" mode.
This is what happened with the original DOCTYPE-switched "standards
mode": many people wanted their pages to use the latest standards, but
found that they ran into bugs in IE6, so they worked around those bugs
with hacks — which relied on simple parsing bugs — to pass
content specifically to IE6. With IE7, many of the simple parsing bugs
that these hacks relied on were fixed, but the bigger underlying
problems that the hacks were working around weren't. The pages thus
broke in IE7. Microsoft's reaction was to make "standards mode" mean
"IE7" mode, and to introduce a new mode ("IE8"). The same will happen
if so many people use "edge" that a significant number of pages break
in IE8 or IE9.
We could encourage everyone to actually use this feature, but
actually then we're just contributing to making the syntax legitimate,
and we end up fragmenting the Web, on Microsoft's terms. This is
possibly the worst outcome.
Finally, we could just ignore the feature altogether, continuing to
use JS compatibility libraries for the time being, the same way that
everyone has been doing for years. Authors would also have to support
IE7 anyway, at least for the forseeable future, so it wouldn't be an
additional cost. This would make the experience on IE worse than on
other browsers, but not so much worse as to turn users away from the
sites themselves. Microsoft would find themselves having to support a
number of rendering modes that were not being used, which is not a
cost-effective measure; they would find their market share eroding, as
customers switched to browsers that actually followed standards; and
they would find their developers still asking for standards compliance
that was actually compliant to the specs and interoperable with other
browsers.
Therefore I recommend not including the meta tag, or, if
you are forced to include it, making sure it says "IE=7", even once
IE8 ships. This seems to me to be the best way to show your
support for an open, interoperable Web on the long term.
It will be interesting to see whether IE8 really supports Acid2, since that test page
doesn't include any of the special magic words being proposed
here. Will they hard-code the URI? Will they check every page against
a fingerprint and if it matches the fingerprint of the Acid2 page,
trigger the IE8 quirks mode instead of the IE7 quirks mode?
Third: Well, not really an announcement, but Sjoerd
Visscher pointed out something that for some reason I had never
heard of, and which nobody I have spoken to about it since has ever
heard of: using document.createElement() with the tag
name of an unknown tag causes IE7's HTML parser to start parsing that
tag differently. This piece of information makes building an HTML5
compatibility shim for IE7 far easier than had previously been
assumed. This is probably the most noteworthy news of the day. I don't
really know why I didn't know this before.
Pingbacks:
1
2
3
4
5
6
7
8
9
10
11
12
13