--- On Sun, 5/24/09, Joe Kelsey <[email protected]> wrote:
> From: Joe Kelsey <[email protected]>
> Subject: Re: [uwin-users] strange 'ksh' behavior WRT file extension and 'mv'
> To: "Sergei Steshenko" <[email protected]>
> Cc: [email protected], "David Korn" <[email protected]>
> Date: Sunday, May 24, 2009, 1:58 AM
> Sergei,
>
> Please STOP responding. You do not know ANYTHING
> about the subject. Your input does not serve a useful
> purpose. Please GIVE UP on the idea of cross-compiling
> windoze executables on Linux. That is a physical
> impossibility due to Microsoft restrictions on dll
> usage. You CANNOT create windoze executables on a
> Linux platform no matter what you do because Microsoft
> FORBIDS it in their license.
> Sergei Steshenko wrote:
> >
> > And I think the intent is wrong.
> >
> > If there are both foo.exe and foo I do not want to
> have doubts.
> >
> > And the type of OS _can_ be determined, so the
> assumption that foo is expanded to foo.exe should be made
> only in windows 95, 98, me, or not at all - the mentioned
> OSes are not relevant.
> >
> There is no reality in which this is an issue. UWIN
> runs on windoze and nothing else. There is no
> possibility for any other kind of mapping function.
> You cannot decree something else just because you are
> confused. Go away and think of some other project to
> do. I suggest that you first learn AST, then learn
> UWIN and get real familiar with Microsoft restrictions on
> usage of their property.
>
> /Joe
>
>
The project name - UWIN - means UNUX on/for Windows.
In UNIX files names are taken "as is", with no interpretation.
So, the behavior should be the same as under UNIX. There might be an
exception for calling a program for execution.
At all, I would introduce an environment variable, say,
UWIN_SHOW_FILE_EXT
, and would change file name mapping functions to behave according to the
environment variable setting. In such a manner one would be able to choose
pure UNIX behavior or "hidden executable extension" Windows behavior.
Windows programs know nothing about the environment variable, so their
behavior won't be affected.
And I read a lot of articles about UWIN, including a quite interesting
article by David Korn (in PDF, don't remember its name at the moment)
describing UWIN internals and design choices; extent of UNIX behavior
emulation, etc.
Regarding creating prohibition to create Windows executables on other
platforms - even if it is true, it is not a problem.
During the development phase I can use cross-compilation, and perform
simple checks using WINE, and when I think the code is good enough for
real testing, I can recompile it from sources on Windows and, if I need to
ship it, ship the compiled Windows version.
Anyway, this prohibition (if it exists) is widely and publicly ignored -
see, for example:
http://gnuwin32.sourceforge.net/ -> Compilation ->
http://gnuwin32.sourceforge.net/compile.html -> (look at the bottom) ->
http://www.profv.de/mingw_cross_env/ ->
http://www.profv.de/mingw_cross_env/README.html ;
and another one:
http://www.sandroid.org/imcross/ .
I tried http://www.profv.de/mingw_cross_env , works fine for my simple
tasks.
And I do not decree, I'm telling about broken expectation based on previous
experience with UNIXish systems.
Regards,
Sergei.
_______________________________________________
uwin-users mailing list
[email protected]
https://mailman.research.att.com/mailman/listinfo/uwin-users