[cgiapp] Re: Unexpected variable persistence with CAD / CADS

[George Hartzell]

Richard Jones writes:
 > George Hartzell wrote:
 > > You didn't actually provide a broken example, so I cobbled one
 > > together that I think does what you're describing, but it doesn't
 > > exhibit the problems you've described.
 > >
 > > I've stuck a shar archive of a little app on the end of this message.
 > >
 > >   http://localhost:8080/foo/update/2 
 > >
 > > followed by
 > >
 > >   http://localhost:8080/foo/new_record 
 > >
 > > does not seem to have a stuck id value.
 > >
 > > Does it have the problem on your system?
 > >
 > > If not, can you figure out how to make it have the problem?
 > 
 > Yes!! Just put a PARAMS entry in Dispatch:
 > 
 > sub dispatch_args {
 >   return {
 >     table   => [
 >       ':app/:rm' => { },
 >       ':app/:rm/:id' => { },
 >     ],
 >     args_to_new => {
 >       PARAMS => { foo => 'bar' }, # <= comment to change behaviour
 >     },
 >   };
 > }
 > 
 > Your example app. will now exhibit the described behaviour just using:
 > 
 > http://localhost:8080/foo/new_record
 > http://localhost:8080/foo/new_record/123
 > http://localhost:8080/foo/new_record
 > 
 > Even an empty PARAMS arg (PARAMS => {}) switches on variable 
 > persistence. See also the debug output to console to see the id param 
 > presence in every request.
 > 
 > Not sure if this is the intended behaviour, but presumably means it's 
 > not possible to use the PARAMS arg in args_to_new - at least not without 
 > being very careful and fully aware of the consequences.
 > [...]

Yeah, args_to_new and CAD::Server have issues.  You may remember the
last time they bit you:

  http://lists.openlib.org/pipermail/cgiapp/2008q3/000656.html
and
  http://lists.openlib.org/pipermail/cgiapp/2008q3/000695.html

I think that the shared thought of that thread was that CA::Server
should be fluffed up a bit to include some functionality that folks
wanted and then CAD::Server should be retired.

To that end I fixed what folks were asking for, patch thread starting
here (with a bit of self-commentary in the following posts): 

  http://lists.openlib.org/pipermail/cgiapp/2008q3/000695.html

and we're waiting for a new CA::Server release w/ bated breath.

g.