[Xastir-dev] acinclude.m4:XASTIR_DETECT_DEVICES

[Jack Twilley]

WARNING: Unsanitized content follows.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

>>>>> "Curt" == Curt Mills <archer at eskimo.com> writes:

Jack> No, that's a bad idea.  It goes against the grain of autoconf
Jack> and good software practices.  The code is very simple as it is:
Jack> Test for the presence of /dev/ttyS0 and use it if it's there.
Jack> Test for the presence of another popular serial file name and
Jack> use *that* if it's there.  I don't see any gain in changing it
Jack> in any way other than adding in future popular serial file
Jack> names.

Curt> Aren't you forgetting why we're revisiting the issue?  When
Curt> configure checks for "com1" on some WinXP boxes (and perhaps
Curt> Win2k boxes, I forget), it hangs.

I'm remembering why we're revisiting the issue, I'm just making sure
any fix we install is consistent with our goals. :-)  Strictly
speaking, I should have said "adding in future or subtracting
obsolete" instead of just "adding in future".

Curt> If those boxes hang on the "com1" detect, they may also hang on
Curt> a "/dev/ttyS0" detect.  Only one way to find out: Have someone
Curt> with the hang program try new code to see if it flys.  That's
Curt> configure hang is why I changed the code so that we no longer
Curt> try to detect "com1", but we look for another Cygwin signature,
Curt> the "/proc/registry" directory presence.

That is a unique identifier for Cygwin, but it turns out that we don't
need a unique identifier for Cygwin in this case.  Here's the
changelog for the appropriate part of the Cygwin code:

2001-10-29  Christopher Faylor  <cgf at redhat.com>
        
        * fhandler.h (fhandler_serial::fhandler_serial): Change to only accept
        unit argument.
        * fhandler_serial.cc (fhandler_serial::fhandler_serial): Ditto.
        (fhandler_serial::open): Avoid else when previous clause is a return().
        * path.cc (get_devn): Alias /dev/ttyS0 -> /dev/com1, etc.
        (get_device_number): Reallow standalone "com1" as a valid name for   
        /dev/com1.

This tells me we can completely strip all references to "com1" from
our code.  Does this seem reasonable to you?

Jack.
- -- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (FreeBSD)

iD8DBQFAYLf5GPFSfAB/ezgRAkHWAJ4njdfacSmUIfGqfhtuV+HUkn0GIwCfdAZj
XCbT0/UCDIjUzSCkWEvghTU=
=ekNj
-----END PGP SIGNATURE-----