RE: [css3-*] Defining support for the inherit keyword

[Boris Zbarsky:]
> 
> On 10/28/11 11:20 AM, Sylvain Galineau wrote:
> > Have there been complaints about it?
> 
> Yes.  There have been several cases of people forgetting to put it on a
> property they're adding to a module and then confusion ensuing all around.
> 
> Also, as an implementor, anything that simplifies the text in the "allowed
> value" line is good.  The text there can get pretty hairy as it is; if it
> always had to include a few extra "||" and parens and such it would be
> even less readable than it is now.

I'm open to making things more readable; but I don't think readable means
incomplete and inconsistent. It would be silly to repeat the same long 
series of things every time if we can shorten it without losing valid information. 

> 
> > Assuming that spec readers -
> > especially the implementers among them - prefer information to be
> > distributed across more documents than necessary
> 
> I don't see how this is any worse than CSS core syntax being defined in a
> separate document from all the various CSS3 modules.

So because there is already some confusion, we shouldn't fix any? This kind
of argument is generally unpersuasive. And, by the way, I'm not claiming this 
issue is more important than others in the same class. I've just happened to
run into this one lately.

> 
> In fact, just adding this bit to the core syntax document or values and
> units (which all other modules depend on anyway) would not increase the
> number of documents involved.

> 
> So this particular implementor would vastly prefer this be specified
> centrally and that allowed value lines be made as readable as possible,
> per above.

Sure. I have no issue with that, as long as the dependency from values/types 
to their central definition is explicit, visible and unambiguous. And whatever
the solution, I would  like us to be consistent. As things are, we have: 

1. One CSS3 module that says initial and inherit are global values
2. A number of CSS3 modules which, for historical and editorial reasons, 
explicitly list inherit as a valid value *without* listing  initial; which, 
given a central definition, could be interpreted as overriding css3-values 
and excluding initial.
3. A number of CSS3 modules that say nothing at all. 

I don't understand how this increasingly sloppy hodge-podge adds to clarity.
Again, yes, there are bigger issues on the overall clarity scale but I doubt
the time it takes to fix the smaller ones is a real obstacle in addressing them.

Received on Friday, 28 October 2011 16:35:54 UTC