FreeTDS Development Group

["James K. Lowden" <jklowden AT freetds.org>]

Reid, Roger L. wrote:
> Sorry, [...]

No problem.  :-) 
 
> Yes, that gets the message out in the two source files I tried it on.

Good!  

> In dblib/bcp.c I put it right after  #include <assert.h> and the warning
> went away.   
> 
> Order matters.   I tried it at the top of tds/mem.c (another of the many
> files affected) and changed the warning to ../../include/config.h:393:1:
> warning: "_FILE_OFFSET_BITS" redefined

That's odd.  

> But oce I moved it under the assert.h in mem.c it was happy.

Well, the question in my mind is why arpa/inet.h isn't required by other
OSes.  Or, more precisely, what assumption are we making that we
shouldn't?   And if I do change all the includes, how do I get them in the
right order?   

> So, I could edit every affectted .c and have a nonstandard source...or
> does this show a more general (maintainable) solution by modifying the
> configure file.

I'm going to think about that, and do some research about the standards. 
It seems like we're missing something, especially because lots of other
people have built FreeTDS on Solaris.  

If you do edit a bunch of files and send us a patch (here), I'll apply it
to CVS HEAD and try it.  If it works, I'll leave it in. 

> And I've stopped using --sybase--compat...thanks

Cool.  

> roger

--jkl

> -----Original Message-----
> From: James K. Lowden [mailto:jklowden AT freetds.org] 
> Subject: Re: [freetds] solaris compile warning on struct in_addr
> 
> 
> Including arpa/inet.h in bcp.c didn't help?  
> 
> --jkl