bug-ncurses
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH] cygwin sharedlib build fixes, terminfo


From: Charles Wilson
Subject: Re: [PATCH] cygwin sharedlib build fixes, terminfo
Date: Sun, 03 Aug 2003 18:45:02 -0400
User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624

Well, as it happens, NONE of my proposed solutions actually work. Modifying the search order in ac-20020210 doesn't help -- because I still get the fullpath:

CXX = /usr/bin/g++

And again, /usr/bin/g++ != g++.

This fullpath issue, as you'll recall from my earlier message, also crops up when you set 'CXX=g++' manually and run confiugre (generated from ac-20020210). The fullpatch gets substituted, even then -- and, of course, it doesn't match the CXX as specified in libtool.

The problem is the non-standard behavior you used (to get around a different nonstandard behavior...):

       if test "$CXX" = "g++" ; then
                AC_PATH_PROG(CXX,g++)
        fi

which resets CXX to the full path as found by AC_PATH_PROG. Thus, if and only if the user is using g++, force CXX to contain the full path to g++ no matter what the user desires. But, if the user is using ANY other cplusplus compiler, allow 'bare' unqualified names... blech!

Perhaps the following?

        # autoconf 2.5x removed the error - by hardcoding it to g++.
        if test "$CXX" = "g++" ; then
                CXXTEMP=$CXX
                AC_PATH_PROG(CXXTEMP,g++)

                if test "$CXXTEMP" = "g++" ; then
                        AC_MSG_WARN(ignoring hardcoded g++)
                        cf_with_cxx=no; CXX=""; GXX="";
                fi
        fi

This would perform the check that you desire and issue the warning when necessary, but without overriding CXX with a full-path-expanded version.

This change, coupled with "some way" of convincing configure to use/find the "proper" cplusplus compiler ('export CXX=g++' works; but I understand your desire for something automatic. Changing the search order works, too), gets the Makefile "right".

--
Chuck







reply via email to

[Prev in Thread] Current Thread [Next in Thread]