Disambiguating @summary from a long descriptor

As you know, I have been involved in the @summary debate for some time.
I wish to offer the following as some thoughts to clarify where I now
stand on the issue and to also clear up any confusion that I now realise
I may have contributed to regarding @summary.

Please note these points do not represent the view of PF although we
will be discussing it shortly.

To start, I thought of @summary pretty much as a long descriptor for
tables, and there were aspects of its use in this capacity I saw as very
useful - but I realise it may not be wise to /continue/ with this view
of @summary - as a long descriptor - if we are to progress the current
specification, as it has some severe limitations.

In short, examining @summary in the HTML 4 spec would lead one to
believe that it is a long descriptor. It is may not be explicitly
defined as such but the implication is certainly there.  The DTD
definition found in HTML 4 is "purpose/structure for speech output" so
the idea of @summary as a long descriptor is certainly easy to conflate
with this definition. However, and more importantly to /progress/ this
issue, I realise now that @summary should be used to only describe the
structure of a complex data table alone due to some of the limitations I
mention below.

1) There has been talk of exposing @summary in the browser etc or its
benefits to other user groups. As it stands this may not be the best
avenue to go down due to progress with ARIA due to the tight binding
that there is currently between the <table> element and @summary.
@summary is therefore only for the <table> element and is an /attribute/
mostly useful for blind/VIP who are screen reader users and cannot
visually make associations between data cells in a complex data table.
There are looser more generic descriptors that could be used as long
descriptors instead.

2) Because it is only an attribute with the limited capability that
entails - @summary is not actually suitable as a proper long descriptor
due to its inability to provide the semantic backbone that a true rich
content description would actually need. To use it as such would
probably be a retrograde step.

3) It may just be a one trick pony. It does that trick well, but in this
instance we may be best to leave it as it is - disambiguate it from this
notion of it as a long descriptor and promote aria-describedby or
similar suitable mechanisms. This could also help to produce a more
useful true long descriptor which could be used by other groups if
substantive research indicates that this is the case. Currently @summary
may not be suitable for other user groups.

4) Finally it should be appraised just for what it is - a mechanism for
describing the relationships between data cells in complex tables that
is useful for blind and VIP and we need to explicitly keep it as such. I
would be happy to keep @summary in the HTML 5 spec and  just remove the
word purpose from "purpose/structure for speech output" and advocate for
the better long descriptor. This would support backwards compatibility
with older UAs and as the newer long descriptor gained traction that
could be implemented by AT vendors etc.

I hope this does help to clear up a couple of things and I am sure we
can come to a better solution.

Cheers

Josh

[1] http://www.w3.org/TR/html4/struct/tables.html#h-11.1.1

Received on Wednesday, 10 June 2009 09:03:30 UTC