I think you'll find that --ignore-exit is indeed ignoring the exit value; however, it is not ignoring failed tests.

In "... our own code for generating the TAP output ...", you'll probably need to implement something along the lines of TODO or SKIP that Test::More provides which, in your code, ignores failed tests where that would be a valid thing to do.

Consider these four runs:

$ for i in exit_0_test_todo.t exit_1_test_todo.t; do cat $i; echo "---
+-------------"; prove --failures $i; echo "============"; prove --fai
+lures --ignore-exit $i; echo "============"; done
use strict;
use warnings;

use Test::More tests => 1;

TODO: {
    local $TODO = 'WIP';
    is 1, 0, 'Tickety-boo';
};

exit 0;
----------------
exit_0_test_todo.t .. ok
All tests successful.
Files=1, Tests=1,  0 wallclock secs ( 0.01 usr  0.01 sys +  0.03 cusr 
+ 0.06 csys =  0.12 CPU)
Result: PASS
============
exit_0_test_todo.t .. ok
All tests successful.
Files=1, Tests=1,  0 wallclock secs ( 0.02 usr  0.02 sys +  0.03 cusr 
+ 0.06 csys =  0.12 CPU)
Result: PASS
============
use strict;
use warnings;

use Test::More tests => 1;

TODO: {
    local $TODO = 'WIP';
    is 1, 0, 'Tickety-boo';
};

exit 1;
----------------
exit_1_test_todo.t .. 1/1 # Looks like your test exited with 1 just af
+ter 1.
exit_1_test_todo.t .. Dubious, test returned 1 (wstat 256, 0x100)
All 1 subtests passed

Test Summary Report
-------------------
exit_1_test_todo.t (Wstat: 256 Tests: 1 Failed: 0)
  Non-zero exit status: 1
Files=1, Tests=1,  1 wallclock secs ( 0.03 usr  0.02 sys +  0.03 cusr 
+ 0.06 csys =  0.14 CPU)
Result: FAIL
============
exit_1_test_todo.t .. 1/1 # Looks like your test exited with 1 just af
+ter 1.
exit_1_test_todo.t .. ok
All tests successful.
Files=1, Tests=1,  0 wallclock secs ( 0.03 usr  0.00 sys +  0.06 cusr 
+ 0.03 csys =  0.12 CPU)
Result: PASS
============
[download]