Re: [igraph] Cygwin -> Windows compilation - I get it.

[Mark McClure]

I think I get it now and my previous message was overly pessimistic.
While my executable does still depend on cygigraph-0.dll, the 
important thing is that it *doesn't* depend on cygwin1.dll - and,
therefore, virtually the whole cygwin installation.  Furthermore,
cygigraph-0.dll is really quite small and presumably can be 
distributed with my application.

I'll let you know if I come up with something useful,
Mark


----- Original Message -----
From: Mark McClure <address@hidden>
Date: Tuesday, March 10, 2009 9:58 pm

> Well, I've been trying quite hard for a couple of days now to
> compile an igraph based program in Cygwin that will run on
> Windows.  My ultimate objective is to create Mathematica
> interface to igraph but, if my Windows based colleagues can't
> use it, there's really no point.
> 
> Incidentally, I've been doing this in pure Cygwin - I have no
> MingGW.  Should I?
> 
> At any rate, here's as close as I've got from the top.
> First, cd to src/f2c and edit etime_.c and add:
>   #undef USE_CLOCK
>   #define USE_CLOCK
> just before the #ifndef USE_CLOCK line.  According to Tamas,
> this fixes a compilation error since the MinGW runtime doesn't
> have sys/times.h. Then run the following commands:
>   ./configure --disable-graphml --disable-gmp CFLAGS=-mno-cygwin
>   CXXFLAGS=-mno-cygwin
>   make
>   make install
> 
> The -mno-cygwin flag should build the library without Cygwin
> dependencies.  Then,  we should be able to compile a program with:
>  gcc -mno-cygwin name.c -I/usr/local/include/igraph -L/usr/locallib
> -ligraph -o name  
> 
> Unfortunately, the resulting program still doesn't run.  We can
> see why by examining the DLL dependencies as follows:
>   $ objdump -p automorph.exe | grep "DLL Name"
>         DLL Name: cygigraph-0.dll
>         DLL Name: KERNEL32.dll
>         DLL Name: msvcrt.dll
> 
> The cygigraph-0.dll dependency is still there.
> 
> All for now,
> Mark
> 
> 
>