W3C home > Mailing lists > Public > public-respimg@w3.org > March 2012

RE: -webkit-image-set

From: Ariel <asw3@dsgml.com>
Date: Fri, 23 Mar 2012 16:45:17 -0400 (EDT)
To: Adrian Roselli <Roselli@algonquinstudios.com>
cc: public-respimg@w3.org
Message-ID: <Pine.LNX.4.64.1203231637310.29264@ybpnyubfg.ybpnyqbznva>

A : is just a regular character, so it's an ordinary file with no need 
for server processing - which is important.

I couldn't decide between the two options (regular file, or query 
parameter). I've actually been thinking that maybe we should support both:

<IMG src="foo.png" dpi="70 80 100" dpi-div=".">

The dpi-div character searches the end of the string for the character, 
then replaces it. So the above would become: foo.dpi=80.png

If the char is not in the file, then it's appended:

<IMG src="foo.png" dpi="70 80 100" dpi-div=":"> becomes:

foo.png:dpi=80

If the dpi-div char is a ? then it gets special processing - it becomes a 
url parameter. A ? is added to the end of the src if there is none. If 
there already is a ? then an & is added.

The default would be a .

Another option:

<IMG src="foo.png" dpi="70=foo_70.png 80=foo_80.png 100=foo_100.png">

Basically after each dpi number simply put the desired url. It's more 
flexible, but much more verbose, and also more error prone.

I don't like the highsrc, or 2x things, because they are too limiting, and 
have no provision for the future. (Are we going to have 3.5x at some 
point?)

 	-Ariel

On Fri, 23 Mar 2012, Adrian Roselli wrote:

>> -----Original Message-----
>> From: Ariel [mailto:asw3@dsgml.com]
> [...]
>> In my opinion the image choice should be in the HTML.
>>
>> Specifically I like my idea (obviously :) which is currently option #9 on
>> https://etherpad.mozilla.org/responsive-assets
>
> #9 is interesting, but the query string implies that some sort of 
> server-side processing will happen. Part of me thinks that limits the 
> folks who don't have their own server (or inclination or skill to 
> tinker), which includes so many of the sites on blogs. Those authors may 
> just always serve the large file because they cannot reconfigure how the 
> host behaves.
>
> Even the idea that the browser sends the request with a ":" in the file 
> name still implies some server adjustments to make the filename work.
>
> Unless I am missing something obvious here.
>
> Maybe it just stuffs the number before the extension, pre-pended with a 
> dash/underscore (image_70.jpg)?
>
>
Received on Friday, 23 March 2012 20:45:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 23 March 2012 20:45:46 GMT