I don't know what's your definition of bad practice -- my understanding is
-- be conservative in that which one produces, and liberal in that which
one accepts, which had been widely practiced on the net before HTML even
exists.
Compatibility problem is a fact/life you have to face. You have now
the following choices:
1) ditch the old standard for a complete new one -- temporarily satisfy
your aesthetic and/or logical craving. Result: you need to upgrade all
clients _and_ all servers _and_ all documents -- a support nightmare --
unless you own everything.
2) Make new standard backward compatible without "the bad practice" and
later introduce some client/server format negotiation for old clients.
-- a compromise still causes constant itching in purists' hearts.
Result: if you want to take advantage of new format and still want to
cater older clients, you need to maintain multiple versions of documents
with the same content. _And_ you need to upgrade all existing servers
and all existing clients capable of new format but not the format
negotiation (otherwise, these clients will only get old format) to take
advantage of the negotiation -- still a support nightmare.
3) Make new standard backward compatible _with_ "the bad practice" and in
the mean time proceed with new standard/server/client making. Result:
a less painful transition because new clients can enjoy the benefit of
new format and old clients still get the correct information, especially
in this case, table. No client/server upgrade is _necessary_ just for
using a couple new tags in a document. As more and more clients/servers
get upgraded, "the bad pratice" can fade out.
"The bad practice" acts as a buffer between old and new formats in the
transition period. "The bad practice" provides a mean to reduce transition
cost. "The bad pratice" is considered useful, although some purist may get
heart burns from it.
>future architecture. Inserting all kinds of garbage into the data
^^^^^^^^^^^^^^^^^^^^
Unwarranted generalization.
>stream on the principle that browsers should just ignore what they
>can't handle is a bad idea on the face of it. Please imagine what
>this argument would look like if translated to compiler design,
>database design, or any other kind of computer processing that expects
>structured input.
Weak argument based on a weak analogy -- they don't translate.
__Luke
-- Luke Y. Lu mailto:ylu@mail.utexas.edu/ http://www.utexas.edu/~lyl/