Re: Predefined Class Names Solution

James Graham wrote:

> That rather misses the point of "pave the cowpaths" though -- the idea 
> is to preferentially spec things that have become common practice over 
> things which have not. 

Yes, I do understand that; my problem is that whilst it is easy
to demonstrate that there are $>n$ documents in the wild that
use 'class="copyright"', it is infinitely harder to demonstrate
the intention of the authors in using that construct beyond
the two usages I gave previously (CSS & DOM).  It therefore
seems very unsafe to /assume/ that all usages are in conformance
with the usage now proposed (indeed, earlier posters have shewn
clearly that this is not the case), and since earlier documents
were written at a time when one could use class names with
arbitrary semantics (a time tha still exists, b.t.w.), I cannot
see that ascribing particular semantics to a widely used construct
is actually "paving the cowpath".  As I said earlier, it is more
like deliberately causing the to fork.

If I may adduce a programming analogy, it would be like
suddenly specifying that the open brace /must/ follow
the function name on the same line, and that the
matching close brace /must/ occur on its own in
column-1 one or more lines later.  That may well
ratify 85% of existing practice, but it renders
illegal all documents using the earlier (Algol-68)
convention of indenting the first brace on the
line following the function name, and having
the matching close brace some lines below in the
same column.

Philip Taylor

Received on Monday, 7 May 2007 14:33:56 UTC