Specifications
Why do you include specifications that are not part of HTML5?
HTML5 means different things to different people. You could argue that HTML5 only includes features that are
defined in the W3C HTML5 specification. Or you could argue that it includes every specification, draft or
experimental feature that is added to browsers in the last couple of years. We decided to take the middle
ground and split the test into three parts: the official HTML5 specification, specifications that are related
to HTML5 and some experimental new features that are extensions of HTML5.
Many of the related specifications were at one time part of HTML5. During the development of the specification
they were moved to separate specifications.
But WebGL isn't even a W3C specification!
The W3C isn't the only organization that creates open specifications for the web. The WebGL specification is
published by Kronos, the same group that is also responsible for OpenGL. WebGL is related to HTML5 though and
listed as one of the HTML5 technologies on the W3C HTML5 logo page. The W3C HTML5 specification allows the
canvas element to be extended by new drawing methods and WebGL is one of them.
Why do you test for Web SQL?
The Web SQL specification has been deprecated and replaced by the IndexedDB specification. It is however
still commonly used on mobile phones and at least three vendors have shipped desktop browsers supporting Web SQL. We've decided to include
this specification, but make it a special case. Web SQL is worth 5 points, but only if IndexedDB is not supported.
IndexedDB is worth 10 points. If a browser supports both, only 10 points are awarded. This way browsers
that only included IndexedDB are not penalized, but browsers that only support Web SQL do get some points.
Scoring
What is the maximum number of points you can score?
If a browser passes all tests it would receive the maximum score of 555. Previous
versions of the HTML5test had less tests and therefore also a lower maximum score, such as 160, 300, 450, 475 and 500 points.
Previous versions of this test also awarded bonus points for some features, but as of version 5, we
no longer do.
The scoring seems arbitrary, who decides how many points are awarded?
We decided to award points for each feature depending on how important that feature is for web developers
and how difficult it is to implement that feature. A small and simple feature would be worth less points than a
large and complicated feature. We think this is the most honest way to grade browsers, because otherwise a browser
that only supports the small and simple features would score as high or higher than a browser that went the
extra mile and decided to tackle the big features. But in the end it is based on personal preference,
but I doubt there is a truly objective alternative.
What happens when a browser cheats?
We cannot distinguish between a browser that supports a particular feature and a browser that lies about supporting
that feature. The only way to deal with these situations it to manually confirm the test results. And if a browser
is found to be overly confident about claiming support for certain features we can put that browser on a blacklist.
That means that that even though the browser claims to support a particular feature, we ignore what the browser says and do
not give any points. This is usually just a temporary problem and once the browser has been fixed we will remove
the new version from the blacklist.
Claiming to support a feature which isn't working is not just causing problems for the reliability of the
test results, but there are other real-world problems. For example if you claim to support WebGL, a website may decide
to serve WebGL content. If your browser does not support WebGL, the website may fail in an uncontrollable way. If you
correctly denied support for WebGL, the website may have served alternative content that would work in your browser.
If you claim to support features that you don't, you are breaking the web.
If we find that a browser is structurally lying about which features they support - deliberately or not - we will
usually give a warning to the developers of the browser and if the problem hasn't been fixed in the next version
we will remove the browser from the 'other browser' and 'compare' pages and/or give other penalties. In extreme
cases we may block the browser from showing test results and show a warning instead.
Methods
Why are you using browser sniffing?
Unfortunately, in two very specific cases we are forced to use browser sniffing. The first case is contentEditable
which was not supported on many older mobile devices. Yet almost all mobile browsers claim to support contentEditable.
Fortunately modern mobile devices are starting to support contentEditable, but this left us with a problem. We
cannot reliably detect if a browser has proper support. The only way around this is to use a whitelist of
mobile browsers that do support this feature, otherwise you risk awarding points to mobile browsers that they
do not deserve. The second case is drag and drop, which is also not supported on mobile phones and tablets.
Privacy
What kind of data is collected from visitors?
As this software is managed by TheDoggyBrad Software Lab, privacy is always our priority than any other features or speed. I love privacy and you can prove it yourself, as your web browser's privacy extensions will not detect anything.
Even this software is proprietary, permissions are given to browse the source code via your client's view-source, so you are free to check the source code yourself.
We are not collecting any user and non-user data, when you visit this site. Since the original version of this software by Niels Leenheer is a non-privacy minded software, I am continuing the efforts to remove the lines of codes that are collecting data.