[prev in list] [next in list] [prev in thread] [next in thread] 

List:       busybox
Subject:    Re: inetd for no-MMU in trunk
From:       Denys Vlasenko <vda.linux () googlemail ! com>
Date:       2008-03-29 15:57:33
Message-ID: 200803291657.33627.vda.linux () googlemail ! com
[Download RAW message or body]

On Saturday 29 March 2008 14:10, Bernhard Fischer wrote:
> >Sorry, I thought that it might be not that much
> >mportant. My original patch was to allow 1.9 to
> >compile for no-MMU systems. So it provided
> >obvious benefits.
> >
> >As for 1.10 which resolved the problem otherwise
> >my patch would only remove some code that I think
> 
> If it shrinks an applet without breaking existing functionality,
> then please send a patch. Code shrinkage is always very welcome.

Basically, in case on NOMMU Aleksey proposes to handle
inetd builtins differently - to disable all stream ones.

Current code disables chargen builtin only (both stream
and dgram), and "emulate" two problematic stream ones
by vfork+exec'ing "cat".

Aleksey thinks that execing external programs kinda defeats
the purpose of having builtins. I agree. However, if user
asked for builtins, why should be decide that "we know better"
and break his setup?

By "breaking his setup" I mean that e.g.
"discard stream tcp nowait 0:0 internal" stops working.

We may think that this is bloat but we shall not decide for user.
If uer selected CONFIG_FEATURE_INETD_SUPPORT_BUILTIN_DISCARD=y
and it can be reasonably made to work on NOMMU, why not?

If user doesn't want this bloat, he can simply deselect the option.
--
vda
_______________________________________________
busybox mailing list
busybox@busybox.net
http://busybox.net/cgi-bin/mailman/listinfo/busybox
[prev in list] [next in list] [prev in thread] [next in thread]