RE> Queries could be done with a layer of indirection

Possible Explation of Why URLs Should not Contain a Query:

  1) A URL is a pointer to a resource
  2) The fact that forms data is tacked onto the end of URL's is an
     accident of history. Logically, forms submission data and URL's
     are seperate entities.
  3) Some resources can only be retreived by activating a query on
     some resource container (argument for allowing queries in
     URL's).
  4) Resources of the form in (3) can still be addressed without using
     a query by adding a layer of indirection of the resource
     container.
  5) Forms data should be sent as a message body, not as cruft tacked
     onto the URL.

Thank you, for this, Gavin Nicol. I never thought of that before. To me,
the query part has been the URL (itself). The stuff in front of the query
that advised the Net who could execute the query was a temporary aberration.
With automatic forwarding, most people do not know where their query is
going (finally). As more and more databases go on-line, all of the single
table databases can be handles through these queries (we must still retain
SQL for multi-table queries). So many tools now depend on giving a search
pattern or page number as part of a URL. Individual hypertext tools can 
be programmed to ignore the location part of the URL and to execute the query,
if sufficient local data is cached to do so, and to go to a sequence of
URLs starting with the requested one, if the query cannot be answered.
This is possible because a query has a file component and a string of 
questions (patterns). That file name could be a program to execute. 
If the program or the file which is the object of the query is local, 
there might sometimes be no reason to go outside the local network (provider).
How many tools will no longer be able to hyperlink to a particular
paragraph without queries. (I take it you mean a query to be the 
ANCHOR and the ARGUMENTS in a URL). 

On the other hand, supporting these queries could be a headache for
internationalization. I guess it is time to read how a unicoded URL
will handle an ASCII coded query. The URLs for the new proposed standards
were given at the beginning of this thread.

Received on Thursday, 31 July 1997 15:03:30 UTC