- From: Andy <lordpixel@mac.com>
- Date: Mon, 12 Mar 2001 01:07:43 -0500
- CC: "www-style@w3.org" <www-style@w3.org>
Tantek Celik wrote: > > I am happy to announce the availability (as of a few days ago) of a new > working draft published by the W3C CSS working group: "CSS3 module: Color". > > http://www.w3.org/TR/css3-color > First let me commend you on removing the hugely complicated list of colours from the previous version of this specification. The previous draft defined 210 system colours, but failed to allow one to determine how to style such simple elements as a scrollbar, checkbox, radio button or slider in a way that integrates with the feel of the platform[*]. Despite the inadequacy of the short list of colours in CSS2, vastly increasing the number of colours listed is not a viable solution. With the public unveiling of Windows XP, and the existing demands of the Macintosh and Linux look and feels, its clear that no simple list of colours can accurately reflect even the default look of any of the major platforms; much less what they may look like when customised by the user (themed). Clearly, if CSS is to be able to describe the richness inherent in these looks, it needs a far more powerful mechanism. Therefore its good this list has been frozen - even better if it were explicitly deprecated. I find it a shame that the X11 colour table was included. Though I acknowledge one of the things CSS attempts to do is codify current working practice, I find myself at a loss to discover why this arbitrary colour list actually adds anything of value. The names have little meaning even to native English speakers. These colours are also rather unlikely to render correctly on 16bpp displays. The specification begins by stating it may be judged as appropriate only for 32bpp devices, but this seems to be disingenuous. Clearly it can be useful on much less capable devices, if care is taken to avoid including things that mandate 32 bit colour. So currently the module has: * an arbitrary list of 16 basic colours from HTML - i.e. the Windows ones * an arbitrary list of system colours - i.e. the Windows ones * an arbitrary list of contrarily named colours - i.e. the X11 ones. To this is added flavour, chosen by the user and then somewhat fancifully described as defaulting to the colour of the physical box the user agent runs on. How are most user agent implementors meant to determine this? (given most UAs run on personal computers, not embedded) Ask the user what colour his or her iMac is? What if its Blue Dalmatian? Sarcasm is really out of place here, but again, I find adding one colour for flavour to the list to be inadequate. Mac OS 9 defines a central "flavour" colour and 6 variant shades. Having the central colour available has some utility, but it is of very limited value without knowing the other 6. What is "flavour" on Aqua or Windows XP? One particular shade of blue is clearly inadequate as an attempt to describe the richness of these looks and feels. Even Windows 98 allows 2 dominant theme or "flavour" colours on screen - one for each end of the titlebar. There is much useful information in this module. Opacity, gamma, colour profiles... Mixing in documentation on today's inadequate support for "User Color Preferences" seems of limited value. If CSS is to address look and feel integration, then the working group would better serve the community by waiting until a module which describes how to do this appropriately is complete. Much good work is being done in this area by group members. Codifying more bad practice into the CSS3 standard - even as a stop gap - serves noone in the long run; once in the standard it will be very difficult to remove. I would recommend removing the X11 colours (as irrelevant) and deprecating the system colours (as they are not widely implemented or used, so impact is minimal). AndyT (lordpixel) [*] I admit, my bias is towards general styling of user interface using CSS, rather than styling documents. I'd also comment that the system colours as they stand are similarly biased, including colours for window titlebars and other UI widgets.
Received on Monday, 12 March 2001 01:07:48 UTC