Lists Home |
Date Index |
On Friday 11 February 2005 22:12, Michael Kay wrote:
> > This is the answer: 1) There's no pattern, it's all random;
> > 2) Frans should go do something useful.
> > Right?
> More or less. But there is also an explanation. Given that you're designing
> a language (say XSLT) you've got to decide whether to make an attribute
> name consistent with the style you've chosen for your language or
> consistent with what other people called a similar attribute in a different
> language. It's not an easy choice. In XSLT 2.0, for example, we chose
> "schema-location" on the xsl:import-schema element: we could have used
> "href" for consistency with xsl:include and xsl:import (whose terseness, as
> it happens, is inconsistent with all other attribute names in XSLT), or we
> could have chosen "schemaLocation" for consistency with XML Schema. Given a
> clean slate, I would have chosen "location", on the grounds that "schema-"
> is redundant. You can't be consistent with everyone.
Consistency must be one of the most boring things on this planet. Your
discussion of the dilemmas of achieving consistency in the practice, reminds
me of Jeni's reply, in the thread on xml-schema-dev about element naming
In my opinion, the best naming scheme will differ from markup language
to markup language, and in the end it doesn't matter which naming
scheme you actually use, as long as you use it consistently.
Sometimes aiming for consistency means that you have to adopt the
naming conventions adopted by a markup language that you reuse, or the
database/application from which your markup language is generated.
Another consideration is how the markup language fits in with the
naming schemes used by applications that are used to process the