Re: md5.h and (des.h with linux and openssl)

Joe Greco wrote:
>> POST_HIDENNTPPOSTHOST
>>
>> I defined POST_HIDENNTPPOSTHOST and compiled Diablo
but the compilation
>> threw up errors to do with md5.h. I tried to fix
the problem by using
>> openssl but still had problems. Then I remembered
that Diablo has its
>> own built-in MD5 functionality and that when I had
had a previous
>> problem with md5.h I was recommended to comment it
out. Sure enough,
>> commenting out md5.h meant that Diablo compiled
successfully.
>>
>> That being the case can lines in
diablo-5-CUR-20060921-00/lib/defs.h of
>>
>> #ifdef  POST_HIDENNTPPOSTHOST
>> #include        <md5.h>
>> #endif
>>
>> be removed?
>> If so I think that makes POST_HIDENNTPPOSTHOST
redundant as its not
>> mentioned anywhere else in the code so can be
removed.
> 
> These used to be #defines.  I believe someone made them
runtime
> selectable.  However, what you need to #include is
still a compile-time
> problem, and highly platform dependent.  There's
probably a better way
> to do this.

>From what I can tell all mention of
POST_HIDENNTPPOSTHOST can be removed
from the code and from the INSTALL file because hiding that
header uses
the internal MD5 function and its enabled with the cryptpw
option in
dreader.access. I think that #include <md5.h> has been
left in by
mistake, see below.

RELEASE_NOTES:
2004-07-20
        * dreaderd: Use internal md5 function so that
hashed
          NNTP-Posting-Host headers are enabled with the
cryptpw
          option in dreader.access.

dreaderd/post.c:
        /*
         * Add NNTP-Posting-Host:
         */
        if (*conn->co_Auth.dr_VServerDef->vs_CryptPw)
            MBPrintf(&tmbuf, "NNTP-Posting-Host:
%srn",

NNPHMangle(conn->co_Auth.dr_VServerDef->vs_ClusterName
,
                        conn->co_Auth.dr_Host,

conn->co_Auth.dr_VServerDef->vs_CryptPw));
        else
            MBPrintf(&tmbuf, "NNTP-Posting-Host:
%srn",
conn->co_Auth.dr_Host);

INSTALL:
        #define POST_HIDENNTPPOSTHOST to encode the IP
address of your
        user in the NNTP-Posting-Host: header line.  This is
highly
        recommended.

>> POST_CRYPTXTRACE
>>
>> Another option I defined was POST_CRYPTXTRACE which
includes des.h.
>> Under linux I used openssl to get the DES
functionality. I modified
>> diablo-5-CUR-20060921-00/XMakefile.inc so that for
linux the compile
>> used the headers in /usr/include/openssl and
compiled against openssl as
>> below:
>>
>>  #ifdef __linux__
>> -.set CFLAGS    -g -O2 -Wall -Wstrict-prototypes
"-I$(BD)" $(CDEFINES)
>> -D_FILE_OFFSET_BITS=64
>> -.set CFLAGS    -g -O2 -Wall -Wstrict-prototypes
"-I$(BD)" $(CDEFINES)
>> -D_FILE_OFFSET_BITS=64
>> -.set LFLAGS "-L$(BD)obj" -lfilter
-ldiablo -lm -lz -lrt
>> +.set CFLAGS    -g -O2 -Wall -Wstrict-prototypes
"-I$(BD)" $(CDEFINES)
>> -D_FILE_OFFSET_BITS=64 -I/usr/include/openssl
>> +.set CFLAGS    -g -O2 -Wall -Wstrict-prototypes
"-I$(BD)" $(CDEFINES)
>> -D_FILE_OFFSET_BITS=64 -I/usr/include/openssl
>> +.set LFLAGS "-L$(BD)obj" -lfilter
-ldiablo -lm -lz -lrt -lssl
>>  #endif
>>
>> Don't know if this is correct but is similar to
what is in XMakefile.inc
>> for other OSes.
> 
> Whatever gets it to compile and run on your platform is
correct.  Some of
> us don't have access to a wide variety of platforms for
testing, so we
> kind of take stuff like this on the word of others.

Well, it compiles and runs using that on linux. From looking
at the
XMakefile.inc file I got the impression it had be built up
from
experience and what worked.

>> And should there be 2 CFLAGS lines for linux, which
are identical?
> 
> Probably not.

Removed one of the CFLAGS for linux and it still compiles
okay.

> ... JG

Thanks,
Mike.

_______________________________________________
Diablo-users mailing list
Diablo-usersopenusenet.org
htt
p://www.plig.net/mailman/listinfo/diablo-users

Re: md5.h and (des.h with linux and openssl)

Thread: Re: md5.h and (des.h with linux and openssl)