You are viewing a plain text version of this content. The canonical link for it is here.
Posted to general@jakarta.apache.org by James Todd <ja...@eng.sun.com> on 1999/07/15 03:09:43 UTC

Re: Apache Configuration - was I volunteer!]

<wonderingMyResponseIsn'tReflectedInTheList>
	repost, what's up with this???
</wonderingMyResponseIsn'tReflectedInTheList>

-------- Original Message --------
Subject: Re: Apache Configuration - was I volunteer!
Date: Wed, 14 Jul 1999 17:47:22 -0700
From: James Todd <ja...@eng.sun.com>
Organization: Sun Microsystems
To: general@jakarta.apache.org
References: <DC...@USAMCMS2>
<m3...@tildeslash.com>

Jan-Henrik Haukeland wrote:
> 

> Let me just add this also: To give servlets real momentum, IRL, one
> _must_ get as many ISP's as possible to host them. And since most of
> the bigger ISP's still are using Unix (thank God!), you have to talk
> your average Unix system admin (aka bofh) into hosting servlets. It
> will be bad enough when he sees the XML config files (but he's used to
> sendmail.cf so he'll maybe, just maybe let it pass), but when he see
> the applet's it's game over. And if the Servlet Engine comes with a
> java install-shield (like WebSphere) it's realy game over.
> 

agreed ... what about a framework/architecture whereby
the config info can be accessed and manipulated by a
variety of cooperative means, eg:

	edit config file directly

		stop server
		edit file with your favorite editor (ed to ie)
		start server

	html form/jsp/servlet/config file

		interacts with the very same config file
		mentioned above and does not disrupt
		said process

	applet/jsp/servlet/config file

		some of the same logic used in the
		previous approach can be re-used
		here ... only the "view" changed a
		bit. the key here is the applet
		should interact with the core server
		side config logic/service

	rmi/config file

		remote hooks to server side config
		logic/service

	applet/rmi/config file

		ditto ... just another view

	ldap/config file

		kind of like a broken record at this point
		but the idea is if designed/factored
		appropriately we have compelling options

	serialized object|xml construct/servlet/config file

		this is an interesting way for loose coupling
		(very)remote access (note: add in pki/ssl as
		well)

	there are others ... the key as i see it is that i
	don't think the goal should be one or the other but
	how can this be designed so that options, should it
	be desired, abound.

- james

Re: Apache Configuration - was I volunteer!]

Posted by Alexander Jesse <Al...@Credit-Suisse.ch>.
I would add another setup option
  - make changes using ANY gui to a development/test server
  - reconfig that server
    a) by restart
    b) by a signal-type runtime action
  - create
    c) a script to do the same modifications to another server
    d) an input file for a batch-utility/script to modify other
       servers' config

between a) and b) I prefer b)
between c) and d) I prefer d) (gives me the possibilities to
 write a little utility/script for a new platform. Or even
 a servlet which can read this file (via upload...) and change
 its own servers config -> remote operation)


What does it give me: 
- an easy and controlled way to configure a single server
- the possibilites to mirror these changes to other servers

(might please the Win$$$-type admin as well as the "ed"-loving
 *nix guy)

If I can help in this area: count me in

Alexander Jesse
(former Sys-Prog on tiny to BIG iron)

===========================================================
James Todd wrote:
> 
> <wonderingMyResponseIsn'tReflectedInTheList>
>         repost, what's up with this???
> </wonderingMyResponseIsn'tReflectedInTheList>
> 
> -------- Original Message --------
> Subject: Re: Apache Configuration - was I volunteer!
> Date: Wed, 14 Jul 1999 17:47:22 -0700
> From: James Todd <ja...@eng.sun.com>
> Organization: Sun Microsystems
> To: general@jakarta.apache.org
> References: <DC...@USAMCMS2>
> <m3...@tildeslash.com>
> 
> Jan-Henrik Haukeland wrote:
> >
> 
> > Let me just add this also: To give servlets real momentum, IRL, one
> > _must_ get as many ISP's as possible to host them. And since most of
> > the bigger ISP's still are using Unix (thank God!), you have to talk
> > your average Unix system admin (aka bofh) into hosting servlets. It
> > will be bad enough when he sees the XML config files (but he's used to
> > sendmail.cf so he'll maybe, just maybe let it pass), but when he see
> > the applet's it's game over. And if the Servlet Engine comes with a
> > java install-shield (like WebSphere) it's realy game over.
> >
> 
> agreed ... what about a framework/architecture whereby
> the config info can be accessed and manipulated by a
> variety of cooperative means, eg:
> 
>         edit config file directly
> 
>                 stop server
>                 edit file with your favorite editor (ed to ie)
>                 start server
> 
>         html form/jsp/servlet/config file
> 
>                 interacts with the very same config file
>                 mentioned above and does not disrupt
>                 said process
> 
>         applet/jsp/servlet/config file
> 
>                 some of the same logic used in the
>                 previous approach can be re-used
>                 here ... only the "view" changed a
>                 bit. the key here is the applet
>                 should interact with the core server
>                 side config logic/service
> 
>         rmi/config file
> 
>                 remote hooks to server side config
>                 logic/service
> 
>         applet/rmi/config file
> 
>                 ditto ... just another view
> 
>         ldap/config file
> 
>                 kind of like a broken record at this point
>                 but the idea is if designed/factored
>                 appropriately we have compelling options
> 
>         serialized object|xml construct/servlet/config file
> 
>                 this is an interesting way for loose coupling
>                 (very)remote access (note: add in pki/ssl as
>                 well)
> 
>         there are others ... the key as i see it is that i
>         don't think the goal should be one or the other but
>         how can this be designed so that options, should it
>         be desired, abound.
> 
> - james
>