Re: Annoucement: Local Browser Execution

George Phillips (phillips@cs.ubc.ca)
13 Dec 93 23:46 -0800


Marc says:
>What if there were an attribute to given MIME types, specifiable in
>one's .mailcap, called "receiveoutput" or something similar? This
>would be similar to how mailcaps are also supposed to be able to
>supports "needsterminal" attributes and the like. E.g.:
>
>application/x-csh-exec; my-csh-exec %s; receiveoutput
>
>When the browser fetches a data object of type application/x-csh-exec
>it runs my-csh-exec and, due to the presence of "receiveoutput", pulls
>the output of my-csh-exec back into the browser as an HTTP/1.0
>object...
>
>What am I missing (i.e. what do you want to do with x-exec that that
>wouldn't support)?

That covers the script's output, you're just missing the input.
How would information from the URL get to the script? For example,
in x-exec: you can say "x-exec://program/some/path" and "program"
will get "/some/path" as PATH_INFO. If it wants itself to be
run with "/foo/bar" as PATH_INFO it outputs <A HREF=x-exec:/foo/bar>.
How does the similar event happen with a the MIME type? The
best I can see is that the URL has to bounce off an HTTP server
which outputs the appropriate HTTP/1.0 object with the
application/x-csh-exec output type or somesuch.

In other words, how does the script reference itself in an anchor?

If you can solve that, then I don't think there's any problem
with other stuff like different methods (POST) and extra data
on stdin (for application/x-www-form-urlencoded or whatever).