Re: segmentation fault in test 'ncurses'

[David Ellement]

On 010510, at 16:36:59, Thomas Dickey wrote
> hmm - but the most likely cause of a failure at the indicated point
> (other than bad object code, e.g., from some versions of gcc), is

I'm using gcc 2.7-mentor-96q1p51 on HP-UX 10.20.  I'd been able to
compile ncurses 5.2 + patches to 20010127 without problems


> a problem allocating the 'pairs' array.  I'd check/see if it was
> a null pointer, etc., with the debugger.  Other than that, an
> out-of-range value for its subscript seemed a good guess.

Here's a bit more debugger info:

    #0  0xe670 in attr_test () at ../test/ncurses.c:501
    501                 if (!pairs[pair]) {
    (gdb) bt full
    #0  0xe670 in attr_test () at ../test/ncurses.c:501
            pair = 1792
            row = 2
            normal = 32
            n = 0
            skip = 0
            fg = 7
            bg = 0
            ac = 0
            pairs = (unsigned int *) 0x400134c0
    #1  0x18334 in do_single_test (c=98 'b') at ../test/ncurses.c:3261
    No locals.
    #2  0x18d48 in main (argc=1, argv=0x7b03bdbc) at ../test/ncurses.c:3585
            command = 98
            c = -1
            my_e_param = 1
            default_fg = 7
            default_bg = 0
            assumed_colors = 0
            default_colors = 0
    (gdb) p pairs[pair]
    $1 = 0
    (gdb) quit


Is there something more I should look at?  (Does it make sense that
pair = 1792, when I've got TERM=xterm-256color?)

-- 
David Ellement