Re: [tlug] ACID Tests (was Re: Browser share in Japan?)

["Stephen J. Turnbull" <stephen@example.com>]

Raymond Wan writes:
 > 
 > 
 > On 30/05/11 11:53, Gen Kanai wrote:
 > > The ACID 3 test is flawed for a number of reasons.
 > >
 > > My colleague at Mozilla, Alex Limi, goes into this in detail here:
 > >
 > > http://limi.net/articles/firefox-acid3
 > 
 > 
 > Yes -- actually, I had read about this before on another 
 > page a while ago.  I'm not so sure that this means the ACID 
 > 3 test is "flawed", though.

It does mean it's flawed, in fact.  See the more detailed commentary
at http://weblogs.mozillazine.org/roc/archives/2010/06/not_implementin.html.
(Maybe -- it's a hard slog, as much of the interesting material is in
the comments, but the people commenting mostly need to be whacked with
a cluebat for posting trash.)

In summary, it's flawed because it tests superficial aspects of a
feature, while nobody has an implementation that goes beyond
skin-deep.  Allowing that is bad.  A conformance test should at least
spot-check the hard parts.

I agree with your implied position, though; this is hardly a fatal
flaw in the ACID test, you just have to be very careful when comparing
97% to 100%.

 > It's like those questions on the JLPT or Kanji kentei where 
 > only a small percentage of native Japanese speakers could get.

No, it's not.  The reason is that a statement of conformance to a
standard is supposed to mean "you can use these features to get those
effects", and Font is a *required* part of the SVG standard.  You can
debate whether that was a good idea or not, but that's the standard as
written.  IOW, this is like the difference between 97% and 100% on
kyouiku kanji.  If you miss anything there, you end up looking like
ex-Prime Minister Aso, mispronouncing "mizou" (and creating a fad
among elementary school students, my daughter's friends frequently use
the word!)

 > But...a more worthwhile test would have been one that was 
 > created with feedback from browser developers.

Well ... yes and no.  One suspects that in fact there was feedback
from Webkit and Opera developers, just enough to ensure they get 100%. :-)

However, conformance tests should test what's in the standard, not
what developers think should be in the standard.  They had their
chance; it's not like Mozilla is unrepresented at the standards
conferences.  (Admittedly the paying members have more votes, so bad
standards can be and frequently are rammed down the throats of the
poor users.  Viz, Open Office XML or whatever the standard based on MS
Office docx and xlsx is called.)

In fact, I'm sympathetic[1] with both sides on the issue of whether
SVG Fonts should be a required part of the SVG standard.  I suspect
that SVG Fonts is indeed a bad standard (consider the facts that "*no
browser* supports SVG 1.1 "full" Fonts, and none of them are working
on such support", and IE 9 explicitly disclaims support just like
Mozilla does, quoted from the page linked above).  OTOH, even simple
support has some interesting uses, such as allowing the unique shapes
used in corporate logos to be defined and then included in the
document as text.  Another is the ability to introspect all locally
defined parts of an SVG document with only an SVG editor.  (Both use
cases mentioned in comments, again available at the linked page.)

Footnotes: 
[1]  But not an expert, so don't use my sympathy as evidence for
either side!