W3C home > Mailing lists > Public > www-style@w3.org > January 2006

Background image ideas.

From: Kelly Miller <lightsolphoenix@gmail.com>
Date: Sat, 07 Jan 2006 04:33:16 -0500
Message-ID: <43BF8ADC.8080407@gmail.com>
To: www-style@w3.org

Over the last few days, I've been thinking hard about possible solutions 
to the continuous problems people mention about background-images in 
CSS.  I'm basically submitting this as a series of ideas I had about how 
the background system might be improved.  I don't consider myself 
perfect, or an expert, so these ideas are just opinions and shouldn't be 
considered me lecturing or anything like that.

1)  background-position is too basic as it is right now.  Originally, I 
suggesting augmenting it to allow more control over the position, but I 
had a much better and more backwards-compatible idea: add properties 
that match the position properties (left, right, top, bottom), and 
consider the box created by those 4 properties as the background as it 
is defined in CSS 2.1 (in short, right bottom would be the bottom-right 
of that box, not of the whole element).  Since the default for such a 
situation would be 0 for all 4, not entering them would result in the 
current setup, thus preserving backwards-compatibility.
2)  IMO, adding all the new background properties AND allowing multiple 
backgrounds is majorly complicating the syntax of the background 
shorthand.  I suggest instead of using commas, it may be a better idea 
to consider using a pseudo-element ::background(n), where n is a 
positive integer.  Besides allowing background to remain simple, this 
also allows authors to order their background images in a logical order, 
and also makes it easier for the author to quickly change the order of 
image rendering.  This method would also preserve 
backwards-compatibility, because ::background(0) is the same as 
declaring background on the element itself.  Other interesting possible 
results are using multiple background colors (allowing one to fake 
certain types of gradients in CSS, and saving download time by allowing 
only a portion of the background to be colored a certain color) and 
allowing for a fallback background if multiple backgrounds are not 
supported (since the user agent will understand background on the 
element).  Obviously, ::background(n) should probably be restricted; 
things like float, position and content are useless on such an element, 
so it would have to be restricted similarly to how ::first-line and 
::first-letter are restricted.
3)  An idea I suggested before is to augment background-origin to allow 
using a singular point of reference in multiple elements.  Some example 
additions besides border and padding are root (considered the <html> or 
<body> element in HTML, and the root element in XML), a grid number (for 
the newly proposed grid system; uses the element attached to that grid 
point as the starting source), and maybe something having to do with 
id's.  Obviously, the intent is to avoid using an arbritrary point of 
reference; logically, the source element should be an ancestor of all 
the elements used for background-origins.
4)  I believe besides background-size, another method of doing 
stretching may be useful.  Maybe a property background-dist, which would 
take stretch and tile for values, where tile is CSS 2.1 behavior and 
stretch resizes the image to the size of the bounding box?  I believe 
this would be useful with the coming of vector images such as SVG to the 
web; this would allow using small resized images instead of large images 
in the backgrounds.

-- 
http://www.mozilla.org/products/firefox/ - Get Firefox!
http://www.mozilla.org/products/thunderbird/ - Reclaim Your Inbox!

Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html
Received on Saturday, 7 January 2006 09:33:18 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:42 GMT