Re: POSTing (from mosaic for X)
Rick Troth (troth@rice.edu)
Mon, 1 Aug 1994 11:34:03 -0500 (CDT)
> Related to this, I think it would be a good idea for browsers to support
> POSTing of attributes contained in the same anchor. This would avoid
> passing arbitrarily long command lines to cgi scripts which is unsafe
> as well as hard to read. I've suggested this a couple times before
> in newsgroups., but it didn't catch on, so I try again as a followup to
> your idea.
Having also read Larry's response, in which I think he
made some really excellent recommendations, your wording makes me
think of adding "attributes" or other parameters in the URL itself.
What I mean is, explore/exploit the syntax:
type://[user[:pass]@]host[:port]/[path][;parm;parm;...][?args]
This is going in the other direction, but it works.
We use a subset of this syntax in our interface to our problem
tracking system where some state must be retained from one transaction
to the next. In this case, the path is usually constant, referring
to the same script. [sorry if this is just a rehash for everyone]
The "parm" values could be "var=val" pairs. (they're not for us)
That separator could be ampersand; I've seen commas and semicolons.
> My idea is that with method=POST in an anchor, all the other attributes
> of the anchor would be passed to the cgi script as standard input, just
> as if it were the action of a FORM.
This is worth discussing, whether we embrace it in full
or not. I don't mean to nix it with the alternative method above.
> Dan LaLiberte
> National Center for Supercomputing Applications
>
> liberte@ncsa.uiuc.edu
> (Fight interface copyrights and software patents.
> Join the League for Programming Freedom: lpf@uunet.uu.net)
--
Rick Troth <troth@rice.edu>, Rice University, Information Systems