Re: make 3.81rc1 / MSYS

[Eli Zaretskii]

> Date: Sun, 26 Mar 2006 14:22:22 -0500
> Cc: address@hidden, address@hidden, address@hidden,
>    address@hidden
> From: "Paul D. Smith" <address@hidden>
> 
> The other thing is, if we leave it as-is then this make
> function behaves differently on Windows and UNIX, which doesn't seem
> like a good thing to me.

Well, I'm not sure there's a way around this: "//" has special
semantics on Windows systems (it begins a UNC), whereas on Unix it is
a very old tradition that consecutive slashes are collapsed by system
calls.

> Maybe instead, we should change the platforms where the native
> realpath() doesn't behave properly to use the one built into make
> instead.  Of course the problem with this is that the built in
> realpath() we have today probably doesn't work properly on Windows
> either since it's not DOS-path knowledgeable.

Yes.  And so on second thought I think wed should leave the sources as
they are now, and at most change the test suite to not use "//" on
Windows (or ever, if possible).