This is the mail archive of the mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

SV: Tracing down problem when compiling GCC 3.3 under cygwin PR 10626

>AFAICT, install-sh should never be asked to install an executable which
>lacks an .exe when the executable was built with a .exe extension.  So,
>the fact that sometimes it "works" is a red herring. 

I don't understand what you mean when you say that it is a "red herring".

Here is what I observed:

- The install of gcc fails most of the time under cygwin on the machine 
at work. (This is for sure)
- I have traced it down to install-sh (I could be wrong  here)
- When I launch Emacs and create a shell from within emacs and run
"make install", the install is more likely, but not guaranteed to work. 
(I'm quite confident about this observation).
- I haven't had the oportunity to check if this problem is particular to
my machine.

> As you noted, cp
>has limited support for copying .exe files and mv has none.  It really
>is up to gcc to accommodate that by specifying the .exe extension in all
>cases.  That's the whole point of having an install-sh script.  It's
>supposed to be accommodating of system pecularities.

I have also seen that "ls foo" will list foo.exe. I believe that 
install-sh checks for the existance of "foo" (the source) and 
then believes that "foo" can be used in cp/mv, etc. 

I will convey the message, but from the response I got, I don't 
think the guy who was assigned to the problem had enough knowledge 
about CygWin to resolve the issue.


Unsubscribe info:
Problem reports:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]