My personal needs are for pointers that are compatible with Windows, UNIX and CVS, as a develop on Windows and deploy on UNIX, using CVS to move things around. A simple server-understood pointer file might be the easiest thing to do, and would be easily to extend to support things like server provided environment variables. This is all assuming, of course, that the server will be scanning directories to determine what components to load, as opposed to those components being specified in some other piece of configuration (file or db or whatever), in which case the notion of the pointer is implicit.
-----Original Message----- From: Aaron Mulder [mailto:[EMAIL PROTECTED] Sent: Friday, August 08, 2003 12:21 PM To: [EMAIL PROTECTED] Subject: Re: User Friendliness On Fri, 8 Aug 2003, Erin Mulder wrote: > I'm in favor of that too (both exploded deployments and pointers) > > A lot of the commercial servers let you keep your components anywhere > on the local filesystem -- you just give it a filepath when you first deploy > and then update the files in place to hot deploy. I've used that a lot in > development and it's pretty convenient. Being able to also pull in EJBs > from a separate framework project would be very interesting. > > Another thread talked about needing more than the "drop-in" deployment. > I think this pointer route should definitely be explored. Well, UNIX has symlinks... If we respected shortcuts in Windows, would that be good enough? Aaron
