Not sure if this is WAF 2.0 related - it is not handling at least one "bi-endian" platform correctly now.
Christopher O Cowan
Christopher.O.Cowan at ibm.com
Wed Sep 5 21:43:43 UTC 2018
So, I noticed today that waf config was failing on AIX because it started
passing both the HAVE_LITTLE_ENDIAN and HAVE_BIG_ENDIAN tests. I enjoyed
the error message "Failed endian determination. The PDP-11 is back?",
BTW ;)
The Power Architecture has always been bi-endian, but it seems that it is
now passing the Little Endian check with WAF 2.0.
My understanding is that the newer ARM processors are also bi-endian.
Both platforms default to Big Endian, so I believe the wscript should only
check if both variables are unset, on line 561
buildtools/wafsamba/wscript-559-
buildtools/wafsamba/wscript-560- # Extra sanity check.
buildtools/wafsamba/wscript-561- if conf.CONFIG_SET("HAVE_BIG_ENDIAN")
== conf.CONFIG_SET("HAVE_LITTLE_ENDIAN"):
buildtools/wafsamba/wscript:562: Logs.error("Failed endian
determination. The PDP-11 is back?")
buildtools/wafsamba/wscript-563- sys.exit(1)
buildtools/wafsamba/wscript-564- else:
buildtools/wafsamba/wscript-565- if conf.CONFIG_SET
("HAVE_BIG_ENDIAN"):
buildtools/wafsamba/wscript-566- conf.DEFINE('WORDS_BIGENDIAN',
1)
buildtools/wafsamba/wscript-567-
buildtools/wafsamba/wscript-568- # check if signal() takes a void
function
More information about the samba-technical
mailing list