From: Earnie B. <ear...@ya...> - 2002-11-03 13:57:25
|
Manu (mingw-users) wrote: > Luke Dunstan wrote: > > > >>The problem is that you are using the old compiler, because when using Mingw >>GCC 3.x libmingwex is linked automatically. > > > I understand. The binaries will be linked, but will ./configure find libmingwex? > In that case, with GCC 3.x, I should get: > > $ ./configure > ... Doesn't it look for the function first? If it can't find the function in a normal build it then will search for the function in some famous library. These functions used to exist in different parts of the runtime and caused no problems before. > checking for library containing opendir... -lmingwex > ... > > I mean that even if the compiler can link libmingwex, if ./configure can't find > the library, could it assume that we don't have working dirent functions? > > I had a look to 'libc' from GnuWin32, libiberty and DJGPP sources. While > a few POSIX functions like fork() are implemented as dummy functions, > there's already some useful POSIX functions for Win32. > > What about a POSIXish toolkit for MinGW? > Containing a 'libdir', for example. :) > That's a different question. One I'll neither say yes or no to. Perhaps a include/posix directory where we move the likes of unistd.h to? Perhaps fot the source it exist in mingwex/posix? I'm both for and against the idea. As a maintainer I see the pros and cons of both and they tip the scales equally. In the end though, it's the end user not the developer who will have to live with the choices we make. Should the Win32 user be forced to learn POSIX styles just to use a niffty program that was first written for POSIX. I think not, so my leaning toward more POSIX functionality is swung toward no in favor of the end user. Earnie. |