Re: lynx-dev Lynx2-8-2rel.1 (Compile/Link Problems Under AIX 4.3.1.0)

[dickey]

> On Tue, 15 Jun 1999 address@hidden wrote: 
>  
> > well, it appears that your system defines h_errno something like 
> > h_errno_which() in netdb.h -- my configure test only looks for a variable 
> > by the name of h_errno.  Put an 
> >     #undef h_errno 
> > before the extern int h_errno in HTTCP.c, and it should compile (and link). 
>  
> That's legitimate - my /usr/include/netdb.h has 
>  
>    #if !defined _LIBC || defined _LIBC_REENTRANT 
>    /* Use a macro to access always the thread specific `h_errno' variable.  
> */ 
>    # define h_errno (*__h_errno_location ()) 
>    #endif 
>  
> The equivalent for errno (instead of h_errno) can be found in 
> /usr/include/errnos.h. 
>  
> If the system header files choose to define h_errno this way, we should use 
> it and not override it.  (We may have a similar problem with errno, which 
> would be more important, under some setups). 
>  
> The configure tests shouldn't #undef h_errno, just check that it is 
> there and that it can be assigned to (modifiable lvalue).  Similar for 
> errno, I guess. 

I was trying to avoid making the configure test include the headers since
the network headers themselves may (often do) have odd dependencies.  We're
not defining anything that should use threads - so why would that port
have done this?  (I'm aware of some people who blindly add a -D_REENTRANT
to their compiler definitions, but don't know how we could make their
program build and run properly in spite of their efforts).
   
>    Klaus 


-- 
Thomas E. Dickey
address@hidden
http://www.clark.net/pub/dickey