[PATCH] better ldap,krb5,ads detection (fix bug #125)

Stefan (metze) Metzmacher metze at metzemix.de
Mon Jul 7 12:34:31 GMT 2003

At 14:15 07.07.2003 +0200, Stefan (metze) Metzmacher wrote:
>At 16:18 05.07.2003 -0500, Steve Langasek wrote:
>>On Sat, Jul 05, 2003 at 09:07:47AM +0200, mx at metzemix.de wrote:
>> > > - KRB5LIBS is tainted with the contents of CFLAGS, because
>> > > krb5-config copies the contents of this variable into its output
>> > > - Most of the Kerberos lib checks duplicate the purpose of krb5-config
>> > In my patch the AC_CHECK_LIB_EXT() check only adds the lib when it's not
>> > allready in KRB5_LIBS, so the output of krb5-config will not be extended
>> > with other -lkrb5 ... add the end.
>>Ok, I've tested this, and the resulting lib line (Debian w/ MIT KRB5)
>>looks good.  The only remaining problem we have is with CFLAGS handling
>> > > Do you have plans to address these problems in your patch?  The first
>> > > I can see is not currently addressed (it can be fixed by running
>> > > LIBS="`CFLAGS='' $KRB5_CONFIG --libs` $LIBS"), but it's hard to tell
>> > > from looking at the patch whether you've made a difference for the
>> > > last two issues.  I'd test it myself, but my build tree is currently
>> > > dirty with the contents of my own work in this area. :-)
>> > I don't really understand what's the problem is here?
>>If you mean the CFLAGS issue, it's this: if CFLAGS is set in the
>>environment when running ./configure (it is for us), the contents of
>>CFLAGS are copied into the output of krb5-config.  This gives:
>>KRB5LIBS= -L/usr/lib -gstabs -Wall -O2 -lgssapi_krb5 -lkrb5 -lk5crypto 
>>-lcom_err -lresolv
>>which is clearly wrong; none of the -gstabs -Wall -O2 belong in this
>>list.  The fix is to use the line I mentioned above, which clears the
>>CFLAGS variable when calling krb5-config.
>KRB5_LIBS="$KRB5-CONFIG --libs gssapi"
>will do the trick right?

do we need to clear CPPFLAGS and LDFLAGS also?

Stefan "metze" Metzmacher <metze at metzemix.de> 

More information about the samba-technical mailing list