Re: SO_SNDBUF <wxn2jk62iz.fsf@atog.apk.net>

From: Jeremy Elson (jelson@CIRCLEMUD.ORG)
Date: 11/04/97


This is a minor bug.  Both set_sendbuf() itself and the calls *to*
set_sendbuf() are supposed to be surrounded by #ifdef SO_SNDBUF.

The bug actually was caused by me not being able to make up my mind
between this:

 #ifdef SO_SNDBUF
void set_sendbuf(int desc)
{
  ...
}
 #endif

...

some_other_function()
{
  ...

  #ifdef SO_SNDBUF
  set_sendbuf(d);
  #endif

 ....
}



And this:

void set_sndbuf()
{
 #ifdef SO_SENDBUF
  ...
 #endif
}

some_other_function()
{
  ...

  set_sendbuf(d);

 ....
}


I guess after the third time of switching back and forth I got them
unsynchronized :-).



"d. hall" writes:
>In patch level 12: I was browsing through comm.c, since I was thinking
>about making circle protocol independent or else at least upgrading most of
>it's functions to IPv6 functions, I came across a call to set_sendbuf() in
>init_socket().  set_sendbuf() is only defined if SO_SNDBUF is defined, if
>this socket option doesn't exist (seriously can't think of many legacy
>systems that don't have it), there may be problems.
>
>d.
>
>
>     +------------------------------------------------------------+
>     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
>     | http://democracy.queensu.ca/~fletcher/Circle/list-faq.html |
>     +------------------------------------------------------------+


     +------------------------------------------------------------+
     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     | http://democracy.queensu.ca/~fletcher/Circle/list-faq.html |
     +------------------------------------------------------------+



This archive was generated by hypermail 2b30 : 12/08/00 PST