#3 OS X 10.4 build fails

dan pritts

I've got a stock OS X 10.4.6. I don't have any fink or
darwinports stuff installed.

Build fails with error:

~/src/pwsafe-0.2.0@vfr800% make
g++ -DHAVE_CONFIG_H -I. -I. -I. -lncurses
-I/usr/X11R6/include -L/usr/X11R6/lib -g -O2 -c
pwsafe.cpp: In function `secstring getpw(const char*)':
pwsafe.cpp:1407: error: invalid conversion from 'int
(*)()' to 'int (*)(const char*, int)'
pwsafe.cpp: In function `secstring gettxt(const char*,
const secstring&)':
pwsafe.cpp:1433: error: invalid conversion from 'int
(*)()' to 'int (*)(const char*, int)'
make: *** [pwsafe.o] Error 1

I'm not a c++ programmer so i'm way behind the curve in
debugging this. I did notice that the two lines of
code in the error message were readline-related so i
reconfigured --without-readline and it builds OK.

I'd like to have readline support if possible. I am
guessing apple includes a crufty old version of
readline but i don't know how to get the version out of
the library.


  • Nicolas Dade

    Nicolas Dade - 2006-12-03
    • assigned_to: nobody --> ndade
  • Nicolas Dade

    Nicolas Dade - 2006-12-03

    Logged In: YES
    Originator: NO

    Indeed it's a crusty old readline.h which is the problem. the readline.h is written for C, not C++, and isn't written in proper C89 so it does not declare the arguments to the callback functions (the empty "()"). C++ cannot handle that.

    I have workarounds for that inside the #ifdef READLINE_H_LACKS_TYPES_FOR_CALLBACKS.

    Adding a #define READLINE_H_LACKS_TYPES_FOR_CALLBACKS at the top of pwsafe.cpp should do it.

  • Nik Friedman

    Nik Friedman - 2007-05-15

    Logged In: YES
    Originator: NO

    I tried adding the #define, as suggested, and the build still failed with the same error. Same happened when I used a newer version of Readline (provided by Fink).

    pwsafe builds fine with Readline disabled.

  • Nicolas Dade

    Nicolas Dade - 2015-06-14

    I'm going to need the bits of the output of ./configure where I test libreadline (it's near the bottom), and the config.log file that produced to debug this further. (Or a Mac OS 10.4 machine for an hour).

    I'm tracking this in someone else's bug which reported the same build failure



Log in to post a comment.