cygwin-inst-20000304 query

DJ Delorie
Mon Mar 6 17:46:00 GMT 2000

> subdirectories, but the latest version I downloaded has an
> i686-pc-cygwin directory.

Right.  We're working on migrating all the packages to the
i686-pc-cygwin target identifier.

> I have unpacked the cygwin-inst-20000304.tar.gz archive to the
> c:/cygnus/cygwin-b20/H-i586-cygwin32 directory.  Is this the correct
> place for it ?

That is an acceptable place for it.  The H-* directory exists only in
case you want to support multiple *hosts* (like linux or solaris) via
a file server or something.  In the next release, that H-* layer will
go away.  The next subdirectory down from that is for supported
targets (like embedded boards or cross compilers), which is what the
i686-pc-cywin in the snapshots is really referring to.

If it's confusing, don't worry about it.  Just do what works for you.

> What is the best way to go.  I think having the i586-pc-cygwin32 and
> i686-pc-cygwin32 under the H-i586-cygwin32 directory works the best even
> though it is counter intuative.

The H-* directory is for the *host* machine.  The i686-pc-cygwin in
the snapshots is for the *target* machine.  For example, Cygnus may
ship a solaris-hosted cygwin cross compiler, and you'd see a
gnupro/H-sun-solaris2.6/i686-pc-cygwin/include directory tree.

There's also a "build" identifier to complete the set.  So, if I use
my SGI to build a cross compiler that runs under cygwin and creates
programs for an H8/300 eval board, I'd have:

	build	mips-sgi-irix5.3	(where it was built)
	host	i686-pc-cygwin		(where it runs)
	target	h8300-hms		(what it produces)

In your case, you'd have (probably):

	build	i586-pc-linux-gnu
	host	i586-pc-cygwin32
	target	i686-pc-cygwin

although we all know that the last two are really the same thing.

> Is there a new Net release yet ?

We're working on it.

Want to unsubscribe from this list?
Send a message to

More information about the Cygwin mailing list