Bug#663158: Plain "die;" doesn't always propagate exception

[Anthony DeRobertis]

A couple of things:

1. This is an upstream bug, I tested 5.14.2 compiled from upstream
source as well as latest git, both exibit the bug.

2. I've got a git bisect going to find the regression, so far the log
is:

$ git bisect log
# bad: [a35ef416833511da752c4b5b836b7a8915712aab] Remove RC3 marker.
# good: [6d52c880307229a35c23215c596700e716bd7c32] Removing the RC marker from patchlevel.h
git bisect start 'v5.14.0' 'v5.12.0'
# bad: [44ef8c7de01654d0b33d6fd57119d1e622abb5a2] make regen; make regen_perly
git bisect bad 44ef8c7de01654d0b33d6fd57119d1e622abb5a2
# skip: [b3364d989906ac4625608c0ff3bcfdad7e989cb7] perldelta up to be34c38
git bisect skip b3364d989906ac4625608c0ff3bcfdad7e989cb7
# bad: [3bc9c68be9b73fccbaf3154e7c08c09e241a3be6] Disable regeneration of regcharclass.h on win32
git bisect bad 3bc9c68be9b73fccbaf3154e7c08c09e241a3be6
# bad: [1db4d19556a36b5a8e8604c1e7656999ebc7732b] RT 75902: Add prototypes for tie() and untie() to allow overloading
git bisect bad 1db4d19556a36b5a8e8604c1e7656999ebc7732b
# bad: [9e5bbba0de25c01ae9355c7a97e237602a37e9f3] perlunifaq, uniintro: fix for 80 col display
git bisect bad 9e5bbba0de25c01ae9355c7a97e237602a37e9f3
# good: [83aa740ef4e0ec94b144304b034baf00e5d12723] consting in lex_stuff_pvn
git bisect good 83aa740ef4e0ec94b144304b034baf00e5d12723
# good: [992b236353d3493be5063165567838e96570135a] Don't use a C++ keyword as a variable name ("new").
git bisect good 992b236353d3493be5063165567838e96570135a
# bad: [99782e35be86d92df5daa0659d4cb2351d4a36b9] Merge remote branch 'zefram/zefram/reliable_exception' into blead
git bisect bad 99782e35be86d92df5daa0659d4cb2351d4a36b9
# good: [4f0556e9541e0e65f0abc5f4f6caf1f16ddf2dcd] Permit array assignment to steal temps and copy shared hash key scalars.
git bisect good 4f0556e9541e0e65f0abc5f4f6caf1f16ddf2dcd
# good: [6550424581c83583e72aa799a12d2a2e68d9e5f1] For SAVEt_BOOL, save the value with the type.
git bisect good 6550424581c83583e72aa799a12d2a2e68d9e5f1

It should finish up after I leave, I'll report back the results
tomorrow.

--