[whatwg] The src-N proposal

[Tab Atkins Jr.]

On Fri, Nov 8, 2013 at 10:30 AM, Ian Hickson <ian at hixie.ch> wrote:
> On Fri, 8 Nov 2013, Yoav Weiss wrote:
>> Following some discussion regarding the src-N proposal[1] on the
>> blink-dev[2] and webkit-dev[3][4] mailing lists, I think it's a good
>> idea to move the discussion to a vendor neutral list, where all vendors
>> can freely participate.
>
> FWIW, my view is that src-* tries to solve too many problems, and as a
> result is unmanageably complex.
>
> If the art direction use cases that srcset="" supports aren't sufficiently
> broad to be worh it, I would be more in favour of moving HTML away from
> supporting the art direction use cases natively at all rather than coming
> up with even more elaborate mechanisms.

I don't understand this paragraph.  The objection to srcset has never
been that it's not "broad enough", it's that it's too verbose for
reasonable cases.  It's only short enough to be usable in some types
of simple cases.  In more complex cases, or in other types of simple
cases (namely, variable-width images), it gets very verbose if you
want to do it *right*. This, of course, just means that people wont'
do it right, or will defer to JS to do it right for them, which
defeats a lot of the advantages of having a syntax for it in the first
place.

Another objection is that it reinvents MQs with a new syntax, which
has been a persistent complaint since the thing was first introduced.
People don't like the w/h descriptors, shrug.

src-N solves exactly three problems, the three problems that were
identified by the RICG in their use-case analysis: Multiple Densities,
Art Direction, and Variable-Sized Images (combined with the previous
two).  srcset only reasonable solves the first two; my blog post at
<http://www.xanthir.com/b4Su0> shows how src-N's solution is quite
simple and reasonable, while srcset is crazypants verbose *even if you
limit it to only addressing 1x/2x resolutions*.

~TJ