Re: [Caml-list] Re: portable truncate

[Florent Monnier <monnier.florent@gmail.com>]

Hi David,

2013/07/27, David Allsopp wrote:
[...]
>> 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
>
> That page is slightly out of date as it doesn't refer to the 64 bit MinGW
> port :o)

We are then in a RTFM vs. WTFM situation :o)

We may then take this opportunity to import an up to date information
on ocaml.org


>> 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.
>
> Correct
>
>> 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.
>
> Native in this context means "not Cygwin", indeed.

Except that MinGW comes with Cygwin, and so is also automatically
updated from its package management.

As a result it may not be that obvious for beginners and non-experts
if it's not written and explained explicitly.

Also on ocaml.org this ocaml variant is called:
  "Cygwin-based native Win32 port"

Which also doesn't help to know if mingw variant should be considered
as a cygwin variant or a win32 variant.

> I've done some limited Googling, so may be barking up the wrong tree. I
> believe that in Microsoft C world, the function ftrunctate is called _chsize
> (http://msdn.microsoft.com/en-gb/library/aa246710(v=vs.60).aspx) in io.h and
> that MinGW defined a compatible ftruncate in unistd.h a long time ago.

I did similar searches yesterday and already tried _chsize in MyUnix like this:
#define ftruncate _chsize
But it does not work.
The file is not truncated, and _chsize returns -1 which is the error code.


> I expect that the reason for not implementing ftruncate in win32unix
> originally was either that whoever wrote it didn't know about _chsize or at
> the time it didn't work properly (e.g. on Windows 9x). The Windows API
> functions SetEndOfFile/SetFileValidData (which actually implement this) look
> comparatively new (Windows XP+), although Microsoft annoyingly altered the
> "Minimum supported" sections of MSDN a while ago so "minimum" no longer
> means "introduced".

Anyway, MinGW has to use the OS, but is not a product from MS.
I believe that we can do a good use of this "feature".


> Altering unix is relatively easy if you've compiled your own OCaml - edit
> files in otherlibs/win32unix and use make -f Makefile.nt [opt] to recompile.
> You can simply copy the replacement files into your ocaml lib directory
> while experimenting. Patches adding missing functionality to win32unix tend
> to be accepted quite quickly in my experience...

I've made an attempt to compile the ocaml sources under Cygwin using
the MinGW toolchain, but I didn't succeeded.

I can still alter the unix module as I have shown in the previous
email with the module called MyUnix.

But if I want to share my code this won't help me much, I should bring
the fix upstream.
In case I can find a fix (I have good hope, but I'm not sure yet), do
you think I should use mantis, or just send a patch to protz ?

-- 
Regards