Re: Compiling problems ...

From: Dancer <>
Date: Tue, 10 Mar 1998 23:01:06 +1000

Pedro Ribeiro wrote:
> I'v finally tried squid1.2beta ....
> I'v found the fallowing problem while compiling with the options:
> ./configure --prefix=/usr/local/squid
> --enable-snmp --enable-cachemgr-hostname=localhost --enable-icmp --enable-u
> seragent-log --enable-delay-hack --enable-xmalloc-statistics --enable-acltre
> e=splay --enable-dlmalloc --enable-gnuregex --enable-arp-acl
> --enable-arp-acl -> It seems that this option requires some .h's that my
> Slackware Linux don't have:
> #include <net/if_dl.h>
> #include <netinet/if_ether.h>
> Where can i find them ??

If it were me, I would do:

du -a /usr/include | fgrep if_

(and maybe)

du -a /usr/local/include | fgrep if_

(assuming your header files are in a reasonably familiar place)
I don't know what operating system you are using, so that's my best
guess for finding them. Of course, there's a possibility that they are
named differently on your system.

> After removed this option, it still doesn't compiles, now due to some
> problem with the option:
> --enable-acltree=splay
> while compiling ... the gcc reported:
> acl.c: In function `aclParseIpData':
> acl.c:406: structure has no member named `next'
> acl.c: In function `aclParseIpList':
> acl.c:460: structure has no member named `next'
> acl.c: At top level:
> acl.c:2187: warning: `aclDumpWordList' defined but not used
> Taking a look at struct.h i'v seen that the "next" member of the struct
> isn't defined when we use SPLAY trees for acl's ...
> struct _acl_ip_data {
> struct in_addr addr1; /* if addr2 non-zero then its a range */
> struct in_addr addr2;
> struct in_addr mask;
> #ifndef USE_SPLAY_TREE
> acl_ip_data *next;
> #endif
> };
> My guess is that instead of #ifndef USE_SPLAY_TREE , the test should be the
> reverse: #ifdef USE_SPLAY_TREE
> Is that correct ??

Umm. I don't _think_ so. I get the suspicion that the code for
splay-trees and bin-trees in acl.c is a bit funny. It looks like a bit
much cutting and pasting has happened. You can try it though. I'm just
not sure that adding the member to the structure is going to work. I
don't see any other references to it in the error output you supplied,
and that leads me to believe (matching it up with the source code) that
the tree code smells a bit odd.

> I hope that this will contribute for squid development.

Me too :)

> Best Regards.

Please, please, please, shorten your mail .sig. If I wanted your
life-story, I'd ask :)

Received on Tue Mar 10 1998 - 05:15:24 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:39:14 MST