any news about version 3.0? and what's about windows?

I just wasted some hours to get pre2.0 running on my windows machine.
Although I'm using Visual Studio 2008 the cl.exe is not C99 compatible. That'S a well known fact and you seem to use the dynamic arrays :-) well well, I appreciate that.

Nevertheless I tried to compile metis5.0pre2 with Intel 10.1 within the studio with turned on C99 flag. It fails.
Main reason are the not iso conform functions used inside Gk_free.

Here are some steps I tried to get ist running:

- regex.h : beacause I got linke rerrors I changed from preposix to the boost::regex implementaion (that works fine, hurray)

- some other lazy no standard Linux-C-functions I copied from the internet:

* strerror_r (used in gk_strerror)
* strtoll (used in utils.c)
* strtoimax (used in utils.c)

- this typedefs I also need to add:
typedef long long intmax_t;
typedef unsigned long long uintmax_t;

- this functions are unix only functions:
* gk_fexists: S_ISREG is not an iso function
* gk_dexists: S_ISDIR is not an iso function

- in gk_arch.h:
the correct include of crtdefs.h is: #include not #include

- in streamio.c:
#ifdef __MSC__
#define D_SCAN_REPLACEMENT SCNIDX
#define D_PRINT_REPLACEMENT PRIIDX
#else
#define D_SCAN_REPLACEMENT "$1" SCNIDX /*<-error*/
#define D_PRINT_REPLACEMENT "$1" PRIIDX /*<-error*/

#endif

but I still get link errors for:
* vsscanf
* vfscanf
* vscanf

Any idea, if we finally get a windows comatible version? I really think that many people would support you in respect of this matter, because metis is a fantastic very fast mesh partitioning tool. It's state of the art if you visit HPC conferences.

Hoping for the best,
SirAnn