Re: Beginning compile problems

["David Schuetz" <dschuetz@access.digex.net>]

>     So far I've had include problems with port.h (not liking extern defs
of  
>     opendir and cfsetispeed), configure (strftime and socket calls not
found in  
>     [...] 
> 
> Try figuring out a way that ./configure checks the right places
> (libs and include files) for your system and places the needed
> defines into port.h (and please provide a patch for ./configure
> which doesn't break anything else);

I did some of that already.  Included a "dir.h" and "sys.h" (I believe--I'm
at work now) in port.h to fix a couple problems, redefined LOG_PERROR and
STDERR_FILENO, etc.  I couldn't figure out a way to "fix" configure without
basically customizing it for my system (I changed the initial check for
libc to check instead for libsys_s or libMedia_s).  Should configure maybe
check (in addition to "typical" standard places for any given function) all
the libraries in MACHDEPLIBS [or whatever it was called?]  I added the
appropriate libraries to that def. in config.site, but that only gets used
in the make phase, not during configure.

As I said, I've managed to work around those, but I'm still having a
problem with what I'm certain is good code, I just can't tell 'cause I
don't know c++.  How can I trace out what the real problem with the compile
is?  I've even commented out those problem lines [reproduced again below
for those who have joined us late ;-)], but got similar errors later in
RegExDict.  So I'm sure that I've got some basic gcc configuration problem
somewhere, but where?!?

Thanks in advance!


david.


     
>     All those I kinda hacked around, and I figure they're okay, though
they may  
>     be indicative of more serious problems.  Right now I'm stuck on an
error while  
>     compiling RegExPtr.c++, which I've traced (with no help from gcc's
error  
>     message) to three lines in Array.c++:
>     	ITEM * obj = new(ptr) ITEM;
>     		and
>     	ITEM * obj = new(q) ITEM(*p); [twice]
>     Now, I'm comfortable (though by no means an expert) with Objective-C,
but I  
>     can't make heads or tails of this C++ stuff.  What exactly are these
lines  
>     trying to do? [in this case, ITEM turns out to be the class(?)
RegExPtr].  Why  
>     won't it compile? Here's the output of the make:
>     ....
>     = util
>     /usr/local/bin/gcc  -I../../zlib -D__ANSI_CPP__ -I. -I.. -I../../util
>       -I../../util -I/usr/local/include -I../../regex -g -O
>       -x c++ -c ../../util/RegExArray.c++
>     ../../util/RegExArray.c++:
>       In method `void RegExArray::createElements(void *, unsigned int)':
>       ../../util/RegExArray.c++:27:
>         too many arguments for function `void * operator new(long
unsigned int)'
>     ../../util/RegExArray.c++:
>       In method `void RegExArray::copyElements(const void *, void *,
unsigned  
>     int) const':
>       ../../util/RegExArray.c++:27:
>         too many arguments for function `void * operator new(long
unsigned int)'
>       ../../util/RegExArray.c++:27:
>         too many arguments for function `void * operator new(long
unsigned int)'
>     *** Exit 1
>     Stop.
>     
>     
>     I'm running NeXTSTEP 3.2 on motorola hardware, if that helps (which
it  
>     probably won't).
>     
>     I strongly suspect I'm missing something simple somewhere deep in my 

>     configuration or gcc installation, but not having a clue how c++
works, I'm at  
>     a loss as to where to start.  I've been working at this since
Saturday, and  
>     have given up hope of solving it myself.  Any suggestions???
>     
>     
>     Thanks!
>     
>     david.