- From: Ben Boyle <benjamins.boyle@gmail.com>
- Date: Fri, 15 Jun 2007 11:11:11 +1000
- To: "Henrik Dvergsdal" <henrik.dvergsdal@hibo.no>
- Cc: "HTML WG" <public-html@w3.org>
I'm not sure I understand this issue correctly... I'll summarise my understanding so please forgive and correct if I'm off track. I'm not involved in UA development, but I think in this instance you're talking about how a UA that supports HTML5 has to know about all these defaults. It can't rely on parsing the document alone, can't rely on the document having defined all these attributes ... it must know about and use default values where applicable. I think you're proposing that documents be better encapsulated and not have this dependency on default values (which are defined in the spec, thus outside of individual documents). Is that accurate? If so... The concept of default values goes way back into DTDs (and I suspect SGML). It's pretty standard practice to define defaults when defining a language I think. It's expected. What I'm trying to say is, I think you're raising an issue with language design in general, rather than something specific to the HTML. I think HTML5 needs to be consistent with all other specs from W3C and that includes defining default values for attributes. I'm sure this is documented somewhere in W3C about how markup languages are defined, but I haven't been able to find a W3C reference on "how to define a markup language" which I think would answer this question definitively ~:) On 6/15/07, Henrik Dvergsdal <henrik.dvergsdal@hibo.no> wrote: > > > This discussion has become far too philosophical. At the end of the > > day I > > still don't understand what is wrong with the use of the word > > "default". I > > don't think it does any harm. > > Well, I do. The notion of "default state" just isn't compatible with > a conceptual model that maps from concrete instances to states. First > of all, an attribute is always in some state, so a default state will > never apply. Secondly, even if we define some instances as being > "stateless", designating some state as "default" will be an ambiguous > statement, except for very simple cases: > > Let's say we map the keywords "true", "on" and "yes" to the true > state. We then define this as the default state. What value should > the attribute take by default? > > I definitely think there is something to gain here in terms of > readability and precision. For instance, instead of saying: "the > illegal value default is the false state" we can drop the notion of > default state and simply say: "all other values map to the false > state". That statement is more explicit, easier to understand and > it's more precise because it doesn't spoil the conceptual model. > > -- > Henrik > >
Received on Friday, 15 June 2007 01:11:32 UTC