Re: new bug in gplt_x11.c

[Ethan M. <merritt@u.washington.edu>]

On Tuesday 09 March 2010 12:53:16 Allin Cottrell wrote:
> 
> Following recent changes, gplt_x11.c will not compile using gcc
> 4.4.3:
> 
> gplt_x11.c: In function 'preset':
> gplt_x11.c:5025: error: incompatible types when assigning to type
> 'char[256]' from type 'void *'
> 
> The offending line is
> 
>   buffer = NULL;
> 
> 'buffer' having been previously declared as of type char[256].

Ugh. You're right. 
Fortunately it's not too late to fix this in the 4.4.0 source.

> The bad line is reached if XAPPLRESDIR is defined neither in the
> environment nor in the gnuplot source itself, I think.

But how did you trigger this?
So far as I can see, XAPPLRESDIR is defined in the Makefile regardless
of whether its value is NULL.
Are you looking at 4.4 or 4.5?
.../src/Makefile should have a line like this:

AM_CPPFLAGS = -I../term -I$(top_srcdir)/term -DBINDIR=\"$(bindir)\" -DX11_DRIVER_DIR=\"$(X11_DRIVER_DIR)\" -DGNUPLOT_PS_DIR=\"$(GNUPLOT_PS_DIR)\" -DGNUPLOT_JS_DIR=\"$(GNUPLOT_JS_DIR)\" -DGNUPLOT_LUA_DIR=\"$(GNUPLOT_LUA_DIR)\"  -DCONTACT=\"$(EMAIL)\" -DHELPFILE=\"$(HELPFILE)\" -DGNUPLOT_X11=\"$(GNUPLOT_X11)\" -DXAPPLRESDIR=\"$(XAPPLRESDIR)\"

and those flags should be passed to the compiler when gnuplot_x11 is built.
 
> I'd try for a patch but I don't quite understand the logic in that
> part of the code.

The intent was   *buffer = '\0'

Fixing the code is easy, but I'm kind of uneasy that the build process
may have broken somehow so that is can lose the definition of XAPPLRESDIR.

	Ethan