Subject: Re: LibOLE2 and the build problems - An Update
From: Michael D. Pritchett (MPritchett@attglobal.net)
Date: Wed Jun 07 2000 - 22:37:22 CDT
> Appears that it's failing because of an issue with the new makefiles, and
> the 'cygpath' command. The fact that others are seemingly managing to get
> past this point, and past wv, to the goto box, suggests to me that
> something's 'strange' about the CygWin environment on the Tinderbox
> [I'm don't know anything about CygWin though, so I could be wrong].
Sorry I broke Tinderbox compilation. It wan't my plan.
I don't know much about the Tinderbox machine nor the cygwin enviroment, but
it appears to be old. The 'uname -r' command for OS_RELEASE is "4.0".
Currently I am seeing either "2x.x" (B20.1) or "1.1.x" (latest) on my
machine. The makefiles use the fact that B20.1 has a space in the
'uname -r' result and 1.1.x build doesn't. It seems the problem occurs at
A quick solution for Tinderbox in another test:
ifeq ($(OS_RELEASE), 4.0)
sed 's|//[a-zA-Z]/|/|g' | sed 's|/|\\\\|g'
It can be placed after my nested block and will redefine
TRANSFORM_TO_DOS_PATH for the special case of Tinderbox or it can be
incorpated into the nesting structure to avoid the cygpath error.
The 'cygpath' is a command that I have on both my cygwin environments. I
thought it was standard in the cygwin environment. It is only needed under
the one of the 1.1.x builds. If nesting the above block, it can be moved to
just above where it is used and remove the annoying error Tinderbox is
> Jamie -> off to dream of a land where the Tinderbox is always green...
Dreaming of all green Tinderbox too!
Michael D. Pritchett
This archive was generated by hypermail 2b25 : Wed Jun 07 2000 - 20:30:59 CDT