Re: [Caml-list] Re: portable truncate

[Adrien Nader <adrien@notk.org>]

On Fri, Jul 26, 2013, Florent Monnier wrote:
> Hi,
> 
> 2013/07/26, Florent Monnier wrote:
> [...]
> > 1) I thought that the module Unix only contains portable things for
> > Windows, so it's not true?
> 
> This page indeed gives a list of unimplemented functions at the end:
> [1] http://caml.inria.fr/pub/docs/manual-ocaml-4.00/manual036.html
> 
> but it depends which ocaml variants is used.
> 
> This page above says:
> "The Cygwin port of OCaml fully implements all functions from the Unix module.
>  The native Win32 ports implement a subset of them."
> 
> So this page talks about 2 ocaml variants:
> - Cygwin ocaml,
> - and native Win32 ocaml.
> 
> This other page, gives informations for 3 ocanl variants:
> [2] http://caml.inria.fr/ocaml/portability.en.html
> 
> which are
> - native Microsoft
> - native MinGW
> - and Cygwin
> 
> I should probably understand frm this that both
> "native Microsoft" and "native MinGW" are native Win32 ocaml.

Yes, they are.

> I made the error to think that native Win32 port only refers to ocaml
> compiled with MSVC, while it seems that it also include protz's ocaml
> that is a MinGW variant.
> 
> I'm still not sure though that compiling with MSVC or MinGW can be
> considered equivalent.
> And actually they seem not considering this page:
> [2] http://caml.inria.fr/ocaml/portability.en.html

The compiler differs but otherwise the differences are small. The libc
that is used on Windows is Microsoft's (actually called CRT); mingw*
provide some additions to that and apparently, ftruncate is provided.

In otherlibs/win32unix/unix.ml, you have:
  let ftruncate fd len = invalid_arg "Unix.ftruncate not implemented"

And in unistd.h, you have:

  #ifndef FTRUNCATE_DEFINED
  #define FTRUNCATE_DEFINED
  /* This is defined as a real library function to allow autoconf
     to verify its existence. */
  #if !defined(NO_OLDNAMES) || defined(_POSIX)
  int ftruncate(int, off32_t);
  int ftruncate64(int, off64_t);
  int truncate(const char *, off32_t);
  int truncate64(const char *, off64_t);
  #ifndef __CRT__NO_INLINE
  __CRT_INLINE int ftruncate(int __fd, off32_t __length)
  {
    return _chsize (__fd, __length);
  }
  #endif /* !__CRT__NO_INLINE */
  #else
  int ftruncate(int, _off_t);
  int ftruncate64(int, _off64_t);
  int truncate(const char *, _off_t);
  int truncate64(const char *, _off64_t);
  #ifndef __CRT__NO_INLINE
  __CRT_INLINE int ftruncate(int __fd, _off_t __length)
  {
    return _chsize (__fd, __length);
  }
  #endif /* !__CRT__NO_INLINE */
  #endif
  #endif /* FTRUNCATE_DEFINED */

Considering this, you should probably make a patch to add a file
win32unix/ftruncate.c which use the implementation above and replace the
implementation of ftruncate* in win32unix/unix.ml.
I don't know what is the purpose of the __CRT_NO_INLINE define; I've
already asked and will propagate the answer when I get it. It would be
good to have truncate*() too.

(and submit it to the bug tracker :-) )

-- 
Adrien Nader