Re: [Mingw-users] Make configuration for Win32.

[Earnie Boyd]

Please stay on list.

John Cronin wrote:
> 
> Earnie, I'm interested to know how far you got with this.
> 

I didn't try to get far.  I don't have the cycles.

> I tried a few days ago building 3.80rc1 and first ran into the WINDOWS32
> problem, which I fixed with a
> 
> #ifdef _WIN32
> #undef WINDOWS32
> #define WINDOWS32 1
> #endif
> 
> at the top of make.h.
> 

Yuck.  I think that all occurrences of WINDOWS32 should just be modified
to _WIN32 which is the standard documented MACRO.  However, if the
configure process adds this to config.h, I would live with it.

> I then ran into some problems and had to add the files under w32 (as they
> weren't compiled by default) to the top-level, and compile the .c's and add
> them to the link.
> 

That's part of what I want changed.  The configuration should take it
into account.

> I then ran into unresolved symbols (including our old friend bcopy) in the
> link.
> bcopy was easily fixed, and I carried on with a few more defines for
> functions such as
> xmalloc -> malloc
> xstrdup -> malloc;strcpy
> 

These as well, the configuration should handle this.

> and so on, but eventually came up against some such esoteric ones that I
> couldn't find a description anywhere, including
> fatal, error and savestring (obviously searching google for error isn't very
> helpful! although I found some reference to savestring in readline).  They
> weren't defined anywhere in the make sources, the only prototype being an
> extern in make.h, and I couldn't find them anywhere in my mingw (although I
> am still using MinGW-1.1, but with latest w32api and mingw-runtime - my
> connection is a bit slow).
> 

Well, of course we should use the most recent headers and even gcc-3.2;
but

> So if you got this far, did you get these errors or are they due to my
> outdated mingw?
> 

yes, this is why I'm requesting someone look into providing the
necessary fixes.

Earnie.