[whatwg] Allow trailing slash in always-empty HTML5 elements?

[Sam Ruby]

On 11/29/06, Ian Hickson <ian at hixie.ch> wrote:

> I think basically the argument is "it would help people" and the counter
> argument is "it would confuse people". We need evidence to back up these
> arguments so we can make a solid decision. The only relevant data I have
> is that 50% of the web uses trailing slashes, and only 17% uses XHTML.
> This could be used to back up either argument: "clearly people think that
> trailing slashes are allowed, so we should allow them", and "clearly
> people are confused about trailing slashes, so we should get rid of them
> altogether". I don't know which is best.


More data would certainly be better, particularly as this data apparently
doesn't distinguish between "singleton slashes" and "non-singleton slashes",
and I have my intuitions on that matter, but limiting myself to only the
data we have:

I think that it is fair to assume that the majority of the people who
(rightfully or incorrectly) assume that they are using XHTML use trailing
slashes.  Besides, this is the most conservative assumption for the next
point:

It is fair to conclude that that 33% (i.e., 50-17) of those that (in this
case incorrectly) assume that they are producing HTML use trailing slashes.
In Ian's terminology, these people are confused.

The first question I think we can answer fairly conclusively: of those 33%,
how many will become "un-confused" if HTML5 does not permit trailing
slashes?  Hint: the version of HTML they are currently using already doesn't
permit trailing slashes.

The remaining question is how many of the 67% of the 83% (i.e., 55%) of
people who use HTML and don't using trailing slashes would suddenly become
confused (assuming that they aren't already, but just happen to be lucky or
use good tool) if HTML5 were to now given an option that provides them
little, if any value.  This question I don't believe is answerable only by
scanning the documents that they produce.  But I will say that few people
produce documents in a vacuum, and given that 50% of the documents on the
web contain trailing slashes, I dare say that most if not all of them have
already been exposed to such documents.

And like most questions, the mere existence of HTML5 is likely to influence
the answer to the question.

More specifically, imagine two paths:

Path 1: HTML5 permits two authoring syntaxes, and the question as to whether
or not trailing slashes are allowed is forever "it depends".  I continue to
maintain that most people don't understand DOCTYPEs, and will point to the
50% number above as being consistent with that contention.

Path 2: HTML5 permits only one authoring syntax, and permits "XML-style"
notation only to the extent that such syntax wouldn't be interpreted in a
different manner by consumers that only understand HTML.  The documentation
for HTML5 would contain examples of such cases, and any conformance checker
would only point out such examples.

Note: the two paths above are mere thumbnail sketches.  The devil's in the
detail.  For example, technically ' would fall on the wrong side the
argument, but as I can see from the current draft of HTML5, the right
decision was already made in that case.

- Sam Ruby
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.whatwg.org/pipermail/whatwg-whatwg.org/attachments/20061129/6a8f43fc/attachment-0001.htm>