[techs] Re: Issues raised from Techniques teleconference

Michael Cooper writes:
 > 3. User agent compatibility
 > There is a requirement that each Technique state the user agents in which it
 > works. This is useful information, but complex and subject to frequent
 > change. For instance, for some Techniques there may be no current UA
 > support. For others UA support may be partial or variable (e.g., accesskey
 > is supported by some browsers now but it locks out access to menu items that
 > use the same access key - is that "full support"?). There may be other
 > Technique combinations in which a given technique is preferred when target
 > UAs support it (e.g., tabindex to move navigation to the end of the tab
 > cycle) but an alternate Technique may otherwise be needed (e.g., a skip
 > navigation link).
 > Furthermore, User Agent support is likely to change fast, possibly faster
 > than we are likely to update the Techniques documents. There is therefore a
 > desire to externalize the user agent support information, and point to it
 > from the Techniques document. If we do this, we will need to set up a
 > repository of that information. We will additionally need to figure out how
 > Techniques documents can point to this information, when the information is
 > newer than the Techniques doc. One possibility would be that the repository
 > actively point into the Techniques rather than the other way around, but if
 > so how does this affect our definition of this requirement? 

It is also an open question whether we will have the resources to
track down this kind of information. One solution would be to create a
repository in which third parties (including user agent developers)
can provide information on u a support, though of course unless it can
somehow be verified we would have to provide it with appropriate
disclaimers as to accuracy. Of course, references to information
maintained by third parties are an ideal solution.

I suggest we may want to consider reworking the requirement somewhat.
Instead of listing the user agents that do or do not support a
technique, we should only require that for each technique we identify
whether or not the necessary feature support has been implemented at
all, by any user agent, as at the time of publication. Another option
would be to link to third-party information on u a support without
providing any ourselves, or to encourage  u a developers to volunteer
this information (linking to it if they do).

My concern is that we won't have the resources to obtain this
information comprehensively. After all, our resources are stretched to
the limit as it is.

Received on Saturday, 11 January 2003 00:44:20 UTC