Re: [bug-inetutils] [patch] inetd: set environment variables for remote host, ip and local port

[Richard Dawe]

Hi Dirk,

Dirk Jagdmann wrote:
>> I wonder if it would be better to use the same environment variables as
>> "tcpserver" from ucspi-tcp by Dan Bernstein. See:
>>   http://cr.yp.to/ucspi-tcp/environment.html
> 
> A good suggestion. I have revised my patch to use those variables djb
> choose. However while looking over the code I found one flaw with those
> variables, because TCPLOCALHOST and TCPREMOTEHOST require a DNS lookup
> for each connection which can hurt performance seriously. So if
> something like my patch would go into an official inetd release at least
> those to variables should be configurable.

djb's tcpserver has options to enable/disable the DNS look-ups. As you
say, there should be some way of disabling them in your inetd patch.

http://cr.yp.to/ucspi-tcp/tcpserver.html

> Now while I'm not afraid to add some more code, I'd like to hear some
> more comments on this patch and if there would be any chance of being
> included in the mainline inet-utils. If so, what configuration options
> should be used to control setting of those environment vars.
> 
> Whatever I have attached the current state of my patch to this email.
[snip]

I haven't really looked at the inetutils code, so I have no idea if it
fits in the inetutils style or not. I just lurk on this list out of
interest. ;)

One comment I have is that it does not conform to the GNU Coding
Standards -- you need a space between a function and its parameter list,
like this: "func (param);"

Bye, Rich =]

-- 
Richard Dawe [ http://homepages.nildram.co.uk/~phekda/richdawe/ ]

"You can't evaluate a man by logic alone."
  -- McCoy, "I, Mudd", Star Trek