Re: [Sbcl-devel] Pathnames on MS windows - some progress

[David L. <da...@li...>]

Quoting Anton Kovalenko (an...@sw...):
> David Lichteblau <da...@li...> writes:
> 
> > I would vaguely prefer depending on the length in this case.
> >
> >> Any thoughts?
> >
> > It sounds to me like the UNC fix from 1. is mostly independent from the
> > new //?/ approach in 2.-4.  So two separate patches/commits would be
> > convenient when merging.
> 
> While adding long file name support to my branch, I've also backported
> (some) and cherry-picked (some) of the relevant patches into a branch
> based on the /current/ official SBCL trunk. Those commits were made into
> the `lfn-upstream' branch of my github repository:
> <https://github.com/akovalenko/sbcl-win32-threads/commits/lfn-upstream>
[...]

Thank you, I'll take a look at those!

> (I've forgotten how to trick the /original/ SBCL build system into
> cross-compiling with Debian/Linux mingw64, so I'll probably have to
> build /natively/ with MinGW and MSYS, and it seems to require setting
> up MinGW/MSYS on real MS Windows).

Not certain whether it answers your question, but:

SBCL from git has just sufficiently many of your fixes so that it can be
built using Wine, with MSYS/MinGW installed into Wine as usual.

(For others who might not know what that means -- it's documented on:
http://sbcl-internals.cliki.net/Build%20on%20Wine)

If you meant actual use of gcc as a linux->windows cross-compiler --
it's certainly possible, but I've stopped doing it, since "wine sh
make.sh" works just as well and doesn't require changes to the scripts.


d.