Re: case sensitivity in patterns

[David Boyce]

Better not to top-post...

On Thu, Jul 1, 2010 at 9:33 AM, James <address@hidden> wrote:
> Yes (for better or worse) I'm on Windows...

If you have to ask, you know the answer :-)

> So if nothing else, I can get the source and compile it with the flag set?
> (Use make to make make made for Windows? :) )

Yes, you can certainly build your own. It's not even necessary to use
make to make make; there's a bootstrap script as well IIRC. Plus build
help in README.W32.

I would add a couple of things: first, this feels like a flaw in the
GNUWin32 build, in that a Windows version should be assume case
insensitive filesystems. Not my area of expertise, and there may be a
lot of history I don't know, but it might be worth suggesting to them
that they build it that way.

Second, from a brief review of the code it looks like it wouldn't be
hard to change this from a compile-time to a run-time choice using a
special target like .HAVE_CASE_INSENSITIVE_FS. I think the patheq
macro and a few pathname-hashing macros would need to become functions
driven off a boolean and that's about it (aside from testing and
documentation of course :-). This could coexist with the compile flag
which would then specify just the default behavior. However, a 3.82 RC
is due RSN which may mean this is a bad time for updates.

David Boyce