t/op/gc-leaky-box.t, t/op/gc-leaky-call.t: new test failures

[jkeenan]

On a linux/i386 box to which I have some access, today I encountered test failures in t/op/gc-leaky-box.t and gc-leaky-call.t.

Please see attachment. I believe these appeared sometime between last Friday (Apr 08) and today (Apr 11).

Thank you very much.

kid51

[jkeenan]

The same tests are failing for me on my Darwin/PPC box. But, curiously, they are not failing on my Linux/i386 (linode) server where I conduct most of my testing.

t/op/gc-leaky-box.t ....... src/packfile/api.c:866: failed assertion 'pf'
t/op/gc-leaky-box.t ....... No subtests run 
t/op/gc-leaky-call.t ...... src/packfile/api.c:866: failed assertion 'pf'
t/op/gc-leaky-call.t ...... No subtests run 

src/packfile/api.c has been touched by several people in the last few days. Here's the section of the code identified above.

    863 void
    864 Parrot_pf_mark_packfile(PARROT_INTERP, ARGMOD(PackFile * pf))
    865 {
    866     ASSERT_ARGS(Parrot_pf_mark_packfile)
    867 
    868     if (!pf)
    869         return;
    870     else {
    871         /* locate top level dir */
    872         PackFile_Directory * const dir = &pf->directory;
    873 
    874         /* iterate over all dir/segs */
    875         PackFile_map_segments(interp, dir, find_const_iter, NULL);
    876     }
    877 }

[plobsing]

Unable to replicate, however this looks like an overly-zealous assert. Try after 5ebd3eb.

[jkeenan]

Replying to plobsing:

Unable to replicate, however this looks like an overly-zealous assert. Try after 5ebd3eb.

plobsing,

Thanks, that seems to do the trick. make op_tests now passes on my Darwin/PPC box. But I won't be able to check the other linux box until tomorrow.

kid51

[jkeenan]

make op_tests now passes for me on the other Linux box described above. Closing ticket.