[dev-serveez] Re: [Serveez-discuss] Time to think.

[stefan]

On Tue, 6 Feb 2001, Martin Grabmueller wrote:

> > From: stefan <address@hidden>
> > Date: Mon, 5 Feb 2001 23:19:29 +0100 (CET)

Sorry. I missed the right mailing list...

> > 
> > Hello Serveez'ers,
> >   since I started to reorder all the source code you should decide what is
> > the aim of version 0.0.21. Please delete what's not going to be in this
> > and reply it to me.
> 
> Well, personally I would like to see all the points below in the next
> release, but the problem is: is that doable? What points are easily to
> implement and which are not, which are more important than others?
> 
> >  * make protocol detection limits / timeouts configurable
> >  * seperate port configs from servers
> >  * make port config more detailed
> >  * switch to guile, drop sizzle
> >  * implement a dns client
> >  * implement dynamic server loading capabilities (partly prepared)
> >  * make serveez embeddable
> >  * write embedding manual
> >  * document icmp protocol (tunnel server)
> >  * support for raw sockets
> >  * support for ipv6
> >  * write an inetd for serveez (first dynamic loading server ?)
> >  * write a "heavy load coserver" for testing purposes and as example
> >    coserver (like the foo-server we should have a foo-coserver)
> 
> I think that we'll not be able to fulfill the complete wishlist above,
> so I ordered the topics by (my personal) preference:
> 
>  * document icmp protocol (tunnel server)
>  * switch to guile, drop sizzle
>  * make serveez embeddable
>  * write embedding manual
>  * implement dynamic server loading capabilities (partly prepared)
>  * seperate port configs from servers
>  * make port config more detailed
>  * write a "heavy load coserver" for testing purposes and as example
>    coserver (like the foo-server we should have a foo-coserver)
>  * make protocol detection limits / timeouts configurable
>  * implement a dns client
>  * support for raw sockets
>  * write an inetd for serveez (first dynamic loading server ?)
>  * support for ipv6
> 
> Documentation is very important, and the features we already have
> should all be documented. Thus the first point.

Yes. That is work but easy to fulfill. For I have written the code I will
document it either.

> Switching to Guile is next, because I am right now making friends with
> the Guile code and the Guile developers ;) I am willing to assist with
> this.

Ok. Will Guile'ers also accept patches for porting code to Win32 ? I mean
the native MinGW port. If so I'll send to you what was necessary to make
the distribution I made at <http://sourceforge.net/mingwrep/>.

> An embedded Serveez would be great, and our user input says that this
> is a feature very useful to people.
> 
> Dynamic loading is just too cool to let it out, so it's next.

Yes. The configuration code needs a rewrite then in order to protect
Serveez from memory leaks (Raimi: it was not a too good idea to use
permanent allocators). But since we are switching to Guile it is not
really a problem.

> The remaining points are ordered by feeling only, except that ipv6
> comes last.  I think it's important, too, but difficult since Serveez
> is based on ipv4 heavily (IIRC).

What is IIRC ? And: What is the basic thing to do inorder to support IPv6
? I have not yet a slight idea what would be necessary...

Cheers,
        address@hidden