Re: A nit and an addition for the current draft

On Mon, 18 Sep 1995, Bert Bos wrote:

> Bill Perry writes:
>  |Bert Bos writes:
>  |> Steve Grimm writes:
>  |> 
>  |>  |The nit: The text-background attribute has an ambiguous value type.
>  |>  |
>  |>  |<style>
>  |>  |	body: text-background="bluegreen"
>  |>  |</style>
>  |>  |
>  |>  |There's no way for a parser to know if "bluegreen" is a relative URL or a
>  |>  |color name.  Perhaps there should be two attributes for background, with a
>  |>  |defined order of precedence between them.
>  |> 
>  |> The intention is that color names are entered as keywords without
>  |> quotes. The reasoning behind this is, that, presumably, the number of
>  |> color names is small, so they can be entered in the parser's hash table.
>  |
>  |  That seems a particularly poor way to differentiate between the two.  If
>  |a user wants 'readability' they might very well choose to write everything
>  |like:
>  |
>  |body: text-background="red" text-foreground="white" font-style="demi-bold"
>  |
>  |  Ideally, this should `just work right' from the users perspective.
> Why would quotes be `just right'?
> The first will be a (relative) URL's and so that at least works, but
> only if the style designer has provided a definition for "red" on his
> server.
> The other two won't work, because a string is not a valid data type
> for these properties. Are you suggesting that strings should be
> acceptable anywhere a keyword is expected? This may cause confusion
> for the places where the string is to be interpreted as a URL.
> Or do you want separate properties for URL's, such as
> `text-background' vs `text-background-url'? The problem is that this
> may not be very intuitive either. For example:
>   LI: text-background-url = "snowflake"
>   (LI) P: text-background = red
>   OL: text-background = red
>   (OL) P: text-background-url = "snowflake"
> A P inside an LI inherits a background pattern, but it is overridden
> with a background color. A P inside an OL inherits a background color,
> but it is overridden with a background pattern.

I think that background colors and images should not exclude eachother... 
instead, background images with "transparent" regions should show the 
background color in those regions, so that the above CSS could lead to 
white snowflakes on red as the background. This would seem to resolve the 

Benjamin C. W. Sittler
   "I have great confidence in fools -- self confidence my friends call it."
                            --Edgar Allen Poe
mailto:bsittler@nmt.edu                           http://www.nmt.edu/~bsittler/