Re: [css-counter-styles] Applicability of 'pad'

On Sun, Feb 16, 2014 at 9:03 PM, Xidorn Quan <quanxunzhen@gmail.com> wrote:
> It is mentioned in the spec that not all systems are negative-capable,
> or in particular, 'cyclic', 'fixed', and some 'override' are excluded
> from negative-capables. I propose that the same rule should be applied
> to 'pad' descriptor as well.
>
> It is obvious that, any 'cyclic' and 'fixed' has only one symbol in
> its content according to the definition, as a result, any pad width
> larger than 1 only produces a constant prefix for pad part, hence it
> is useless since if anyone wants such effect, he could append the pad
> symbols to the prefix directly.
>
> In addition, 'pad' for counter styles in such systems could generate a
> counterintuitive result. For example, if someone add 'pad' to the
> predefined style 'decimal-leading-zero' which uses 'fixed' system,
> assuming pad is "3 '0'", what he expects might be -099, -009, ...,
> 009, 099, but what he actually gets will be -99, 00-09, ..., 0009, 99,
> which is strange and unacceptable.
>
> In conclusion, I think 'pad' should not be applied for systems which
> is not negative-capable.

While it may not be useful in all cases, and can always be
approximated manually for cyclic and fixed styles, I don't see
anything particularly *wrong* with allowing 'pad' to work with all
styles.  I prefer not to add restrictions just because some
combination doesn't seem useful; as long as it logically works and
there aren't further negative implications, like performance, allowing
all parts of the language to work together produces a simpler language
overall.

On the subject of decimal-leading-zero, I should actually just rewrite
that as as decimal override with "pad: 2 '0';".  That option wasn't
available before!

~TJ

Received on Monday, 17 February 2014 15:51:24 UTC